Articles & guides

Explainers, how-to guides and the occasional news note. The answer comes first, always, then the numbers, diagrams and sources for going deeper.

Answer card: Claude Fable 5.1 keeps input at $10 and output at $50 per million tokens, and drops the cache read rate from $1 to $0.25, a 75% cut.

Dev News

Claude Fable 5.1 Cuts Cache Reads by 75%

Anthropic shipped Claude Fable 5.1 on September 1 with input and output prices unchanged and cache reads cut from $1 to $0.25 per million tokens.

Answer card: Empirik spins out of Sequoia Capital on September 1, 2026 with $21M in seed funding, building an agent that predicts outages by modelling the ripple effects of infrastructure changes.

Sysadmin News

Empirik Raises $21M to Predict Outages Before They Fire

Empirik spun out of Sequoia on September 1 with $21M in seed funding. It reads infrastructure changes before they ship and predicts what they will break.

Answer card: the Linux 7.3 EDAC pull, forgotten during the merge window and accepted as edac_updates_for_v7.3_rc2, stops AMD ATL loading on Zen systems without ECC, adds Intel Starfire and orphans the ThunderX and MPC85XX drivers.

Sysadmin News

Linux 7.3: AMD ATL No Longer Loads Without ECC Memory

The Linux 7.3 EDAC pull landed late. It stops AMD ATL loading on Zen machines without ECC, adds Intel Starfire support and orphans 2 drivers.

Answer card: WinBtrfs 1.10 shipped on September 1 2026, the first release in two and a half years, fixing extent tree corruption when writing to snapshots and adding write support for free space bitmaps and compressed inline files.

Sysadmin News

WinBtrfs 1.10 Ends a Two Year Wait for Btrfs on Windows

WinBtrfs 1.10 arrived on September 1, the first release since March 2024. It fixes extent tree corruption when writing to snapshots, plus a Steam crash and rclone.

Answer card: AMD posted a standalone HDMI Passive VRR patch series for the AMDGPU Linux driver on September 1, 2026, targeting Linux 7.4, enabled by default with an opt out.

Sysadmin News

AMD Splits HDMI Passive VRR Into Its Own Linux 7.4 Series

AMD posted a standalone Passive VRR patch series for the AMDGPU driver on September 1, targeting Linux 7.4. It kills the blank or flicker on VRR transitions.

Answer card: Anthropic committed 35 billion dollars over six years to Lambda for about 350 megawatts in Nueces County, Texas, on a Hut 8 campus where Nvidia is reported to hold the lease.

Network News

Nvidia Holds the Lease in Anthropic's Lambda Cloud Deal

Anthropic signed a six year, 35 billion dollar cloud agreement with Lambda for about 350 MW in Nueces County, Texas. Nvidia is reported to hold the lease.

Answer card: the August 2026 Steam survey puts Linux at 3.90 percent overall, with SteamOS Holo at 21 percent of Linux users and CachyOS at 15.4 percent, up 1.16 points month over month.

Sysadmin News

Linux Slips to 3.9 Percent on Steam as CachyOS Climbs

Valve's August 2026 Steam survey puts Linux at 3.90 percent, down 0.11 from July. SteamOS holds 21 percent of Linux users and CachyOS reaches 15.4 percent.

Answer card: Tripo AI raised 3 billion yuan across Series B and B plus on September 1, 2026, and previewed Tripo P2.0, a 3D native foundation model producing quad topology meshes in seconds.

Dev News

Tripo AI Raises 446 Million, Ships Quad Topology Meshes

Tripo AI raised 3 billion yuan across Series B and B+ on September 1, 2026, and previewed Tripo P2.0, a 3D model that outputs quad topology meshes.

Answer card explaining Happy Eyeballs v3 in Firefox 155: the browser queries the HTTPS DNS record next to AAAA and A, then races address families and HTTP versions in parallel.

Network News

Firefox 155 Races Connections With Happy Eyeballs v3

Firefox 155 enables Happy Eyeballs v3 and QUIC version 2. Connection setup now races DNS record types, address families and HTTP versions at the same time.

Answer card explaining KDE Linux automatic Btrfs snapshots: home directories become subvolumes, Snapper is configured per user, and kio-snapshot exposes past file versions in Dolphin.

Sysadmin News

KDE Linux Makes Every Home Folder a Btrfs Subvolume

KDE Linux now converts user home directories into Btrfs subvolumes, snapshots them automatically with Snapper, and browses those snapshots from inside Dolphin.

Answer card explaining the NVIDIA Vera scheduler patches: firmware capacity differences plus SMT trigger an asymmetric capacity path that treats a half busy core as a full capacity target.

Sysadmin News

NVIDIA Teaches the Linux Scheduler to Pick SMT Siblings

New NVIDIA patches make one SMT sibling the preferred one on Vera Olympus cores, using SD_ASYM_PACKING to extend the asymmetric capacity work posted in March.

Answer card describing The Ur Project: a Rust based distribution building stack from a former Clear Linux architect, aimed at letting others build their own secure distributions.

Sysadmin News

A Clear Linux Architect Starts a New Distro, in Rust

Auke Kok is building The Ur Project, a Rust based system for producing your own Linux distribution. His diagnosis of Clear Linux: outsiders could not contribute.

Answer card explaining the AMD Zen 6 CPPC patches: the kernel hard codes maximum frequencies of 5025, 3524 and 2399 for performance, efficiency and low power cores because ACPI cannot yet report them.

Sysadmin News

Linux Hard Codes Zen 6 Core Frequencies, For Now

AMD kernel patches hard code Zen 6 client core frequencies at 5025, 3524 and 2399 because ACPI has no interface to read them. The new low power core sits at 0.47x.

Answer card describing the Lekuo SouthBridge Max: an AMD B650 chipset on a PCIe 4.0 x4 add-in card providing four NVMe slots, four SATA ports and four USB ports, recognised by Linux as an AMD 600 Series chipset.

Sysadmin News

A PCIe Card With a Whole AMD B650 Chipset On It

The Lekuo SouthBridge Max puts an AMD B650 chipset on a PCIe 4.0 x4 card to add four NVMe slots, four SATA ports and four USB ports, for around 120 dollars.

Answer card summarising Servo 0.5.0: 488 commits from July, a multithreaded 2D canvas worth up to 55 percent better frame rates, text rendering up to 10 times faster, and new Linux aarch64 binaries.

Dev News

Servo 0.5 Threads Its Canvas and Renders DuckDuckGo

Servo 0.5.0 lands 488 commits: multithreaded 2D canvas for up to 55% better frame rates, text rendering up to 10x faster, and Linux aarch64 binaries.

Answer card explaining LUOSession= in systemd 262-rc1: a service unit asks systemd to create a kernel Live Update Orchestrator session, which is kept in the file descriptor store so the workload survives a kexec.

Sysadmin News

systemd 262 Learns Live Update Orchestrator Sessions

systemd 262-rc1 adds LUOSession= so services can hold kernel Live Update Orchestrator sessions, plus randomized restart delays and slice activation limits.

Answer card explaining near live migration in Incus 7.4: incremental ZFS or Btrfs snapshots are sent while the container runs, and it stops only for the final sync.

Sysadmin News

Incus 7.4 Moves Running Containers Without CRIU

Incus 7.4 migrates a running container by shipping incremental ZFS or Btrfs snapshots instead of checkpointing memory. The container only stops for the last sync.

Answer card summarising Linux 7.3-rc1, tagged August 30, 2026: 15,267 commits, about 40.98 million lines of code, and a final release expected in October 2026.

Sysadmin News

Linux 7.3-rc1 Closes the Second Busiest Merge Window

Torvalds tagged 7.3-rc1 on August 30, 2026 after 15,267 commits. The tree is near 41 million lines, and most of that growth is a single register header dump.

Answer card explaining why OpenAI bought tens of thousands of Mac minis and Mac Studios: reinforcement learning on computer use agents, unified memory, and a desktop class machine per environment.

Dev News

OpenAI Bought Tens of Thousands of Macs to Train Agents

The Information reports OpenAI quietly acquired tens of thousands of Mac minis and Mac Studios for reinforcement learning on computer use agents. Anthropic rents its own.

Answer card describing the Pangolin 1.22 AI gateway: clients point at one Pangolin URL, provider keys stay on the gateway, and budgets and usage analytics sit in front of every call.

Network News

Pangolin 1.22 Puts a Gateway in Front of Your Models

Pangolin 1.22 adds an identity aware AI gateway for eight cloud providers plus Ollama and vLLM, and moves browser based SSH, RDP and VNC into the free edition.

Answer card: multikernel Linux v7.0-mk2 boots several independent kernels on one machine with no hypervisor, each owning dedicated CPUs, memory and PCI devices, with a two process context switch at 1.37 microseconds against 3.42 under KVM.

Sysadmin News

Multikernel Linux Runs Real Kernels Side by Side

The first public multikernel Linux tree, v7.0-mk2, boots several independent kernels on one machine with no hypervisor, each owning its own CPUs and memory.

Answer card: OpenAI ends direct model access inside Cursor on November 12, 2026, citing terms of service concerns after the SpaceX acquisition, with OpenAI models carrying about five percent of Cursor traffic.

Dev News

OpenAI Cuts Off Cursor Model Access on November 12

OpenAI will stop serving its models inside Cursor on November 12, 2026, two weeks after SpaceX closed its sixty billion dollar purchase of Anysphere.

Answer card: OpenShot 4.0, released August 30, 2026, replaces the web based timeline with a native Qt one, adds screen and webcam recording, a colour grading workspace with scopes, and local ONNX object masking.

Dev News

OpenShot 4.0 Records, Grades and Drops the Web Timeline

OpenShot 4.0 arrived on August 30, 2026 with built in screen recording, a colour grading workspace, local ONNX object masking and a native Qt timeline.

Answer card summarising the Alibaba Cloud Brazil launch of August 28, 2026: two data centres in Sao Paulo forming the provider's first South American region, taking it to 106 availability zones across 31 regions.

Network News

Alibaba Cloud Opens Its First Region in South America

Two data centres in Sao Paulo give Alibaba Cloud its first region on the continent, taking it to 106 availability zones across 31 regions worldwide.

Answer card summarising reporting on DeepSeek's funding round: about 7.4 billion dollars raised at a 74 billion dollar pre-money valuation, ahead of a possible 2027 listing on Shanghai's STAR Market.

Dev News

DeepSeek Nears $7.4B at a $74B Valuation Before a 2027 IPO

Reporting puts DeepSeek close to raising 50 billion yuan at a 500 billion yuan pre-money valuation, with a Shanghai STAR Market listing pencilled in for 2027.

Answer card summarising kernel.org traffic figures published on August 29, 2026: six million daily requests to the git web interface, 66 percent turned away at the proof of work challenge, and roughly 2 percent estimated to be legitimate.

Sysadmin News

Only About 2% of git.kernel.org Traffic Is a Real Person

Kernel.org's admin published the numbers behind six million daily requests, and almost all of it is automated crawling of commit pages that nobody reads.

Answer card explaining the August 2026 x86 disclosure: new WBINVD skip hints posted to the Linux kernel mailing list on behalf of an unnamed processor vendor that is neither Intel nor AMD.

Sysadmin News

A Third x86 Vendor Is Documenting Opcodes Nobody Can Name

New WBINVD cache skip hints were posted to the Linux kernel list on behalf of a chip vendor that is neither Intel nor AMD, and it is not the first such disclosure.

Answer card describing the fruc_vulkan filter merged to FFmpeg on August 29, 2026, its three optical flow performance levels and its grid sizes from 1 to 8 pixels.

Dev News

FFmpeg Gains NVIDIA Vulkan Frame Interpolation

A new fruc_vulkan filter landed in FFmpeg git on August 29, 2026. It builds interpolated frames from NVIDIA optical flow, with two knobs that decide whether it keeps up.

Answer card summarising the FreeBSD Foundation report on porting AMD ROCm to FreeBSD: patches to AMD's LLVM fork, the ROCm runtimes, drm-kmod and LinuxKPI, with heterogeneous memory management still outstanding.

Sysadmin News

ROCm on FreeBSD Nears Its First Vector Addition

A FreeBSD Foundation internship got AMD's compute stack to link, load and run on FreeBSD. The remaining work is memory management, and that is the expensive part.

Answer card summarising the KVM changes merged for Linux 7.3 on August 25, 2026: kvm_mmu split into three parts, a single gva_walk entry point, and fixes for Xen steal time and the VMX preemption timer.

Sysadmin News

Linux 7.3 Splits the KVM Page Table Walker in Three

The KVM pull for Linux 7.3 is mostly cleanup, and the cleanup is the story: one guest address walker instead of two, plus a fix for guests reporting bogus steal time.

Answer card summarising WIRED's August 28, 2026 report that Meta is testing robots from Watney Robotics, Kinova and ABB at its Altoona, Iowa and New Albany, Ohio data centers on cabling, power cycling and reseating.

Network News

Meta Puts Robots on Cable Swaps in Its Data Centers

Meta is testing machines from Watney, Kinova and ABB on cabling, power cycling and reseating at its Iowa and Ohio campuses. Dense cabling is still winning.

Answer card summarising the Btrfs maintainership change: Chris Mason left Meta on August 28 2026, his MAINTAINERS entry moves from M for maintainer to R for designated reviewer under a kernel.org address, David Sterba of SUSE remains principal maintainer, and Btrfs development continues unchanged.

Sysadmin News

Chris Mason Steps Down as a Btrfs Maintainer

Btrfs author Chris Mason left Meta on August 28 and changed his MAINTAINERS line from M to R. David Sterba stays principal maintainer, and nothing else moves.

Answer card summarising the Debian General Resolution result: Option 5, Responsible Use of Generative AI, won the vote that closed on August 28 2026, with 425 unique voters, no ban on AI tools, contributor responsibility unchanged, and disclosure encouraged but not required.

Dev News

Debian Adopts Responsible Use of Generative AI

Debian's LLM General Resolution closed on August 28. Option 5, Responsible Use of Generative AI, won outright: no ban, no mandate, disclosure encouraged only.

Four Microduck robots from Pollen Robotics in different colourways, each a 25 centimetre bipedal robot with a large single camera eye set in a rounded head, a coloured beak, visible servo motors in the legs, and moulded feet.

Dev News

Hugging Face Opens Orders for a $399 Robot

Pollen Robotics unveiled Microduck on August 27: a 25 cm biped with 15 motors, LiDAR and an Apache 2.0 SDK, at $399, shipping before Christmas 2026.

Answer card summarising the Sovereign Tech Agency investment in Flatpak: 508,640 euros over two years running through the end of 2027, funding four new portals for audio, network, VPN and writing assistance, plus an entitlements system for static permissions and an intents system for declared services.

Sysadmin News

Germany Puts 508,640 Euros Into Flatpak Portals

The Sovereign Tech Agency is funding two years of Flatpak work through the end of 2027: new audio, network, VPN and writing assistance portals, plus entitlements.

Answer card summarising the Andreessen Horowitz Machine Age Fund: 1.1 billion dollars announced on August 28, 2026, targeting data centre chips, memory and networking gear, edge AI hardware and robotics, with named holdings Heron Power, Volta Infra and Unconventional.

Network News

a16z Raises $1.1B for the Metal Under the Models

Andreessen Horowitz announced the Machine Age Fund on August 28, 2026: $1.1 billion aimed at data centre chips, memory, networking gear, edge hardware and robots.

Answer card summarising the Model Hardware Standard research preview: a shared driver layer, device discovery and safety limits letting AI agents operate microscopes, liquid handlers, robotic arms and lasers, opened to selected partners on August 27, 2026.

Dev News

Anthropic Previews a Hardware Standard for AI Agents

Anthropic opened the Model Hardware Standard research preview on August 27, 2026: a driver layer, device discovery and safety limits for lab and factory machines.

Answer card explaining the Linux 7.3 USB Type-C change: the kernel now validates cable identity, passive cable speed and active cable plugs before advertising an alternate mode such as DisplayPort or Thunderbolt.

Sysadmin News

Linux 7.3 Checks the Cable Before It Promises DisplayPort

Linux 7.3 validates USB Type-C cable capabilities before engaging alternate modes, so a cheap passive cable stops producing a DisplayPort state that never works.

Answer card summarising the ReactOS 0.4.16 release: a graphical installer on a combined live and boot image, a Server Core install type, a new ATA driver replacing UniATA, and the Wine fork rebased onto Wine 10.0, with 2,808 commits closing 381 issues.

Sysadmin News

ReactOS 0.4.16 Ships a Graphical Installer and Server Core

ReactOS 0.4.16 arrived on August 29, 2026 after roughly 18 months: a graphical installer, a combined live and boot image, a Server Core mode and Wine 10.0.

Answer card summarising ROCm 10.0: AMD jumped from the ROCm 7.14 production release to version 10.0 on August 27 2026, with ROCm.AI bundling a unified rocm CLI, AMD Skills for coding agents including Claude, Cursor and Codex, and Hyperloom for agentic inference optimisation, on a new six week release cadence.

Dev News

AMD Jumps Straight to ROCm 10.0 and Bets on Agents

AMD shipped ROCm 10.0 on August 27, skipping from 7.14 and putting an agentic toolchain called ROCm.AI at the centre: a unified CLI, AMD Skills and Hyperloom.

Answer card summarising the Linux 7.3 MGLRU change: a patch series from Baolin Wang at Alibaba promotes mapped executable folios after first use so that hot code resists reclaim the way classical LRU already allowed, cutting system time on a memory starved 32 job kernel build from 9248 seconds to 7861 seconds.

Sysadmin News

Linux 7.3 Stops MGLRU Evicting Your Hot Binaries

A patch series merged as the Linux 7.3 window closed makes MGLRU protect mapped executable pages the way classical LRU does. System time fell from 9248s to 7861s.

Answer card summarising the Linux 7.3 block layer changes: RWF_DONTCACHE uncached buffered I/O now supported on block devices thanks to Tal Zussman, which suits databases running on raw devices, alongside dynamic area provisioning for io_uring zero copy receive and NVMe RDMA queue setup parallelisation.

Sysadmin News

Linux 7.3 Brings RWF_DONTCACHE to Block Devices

Uncached buffered I/O now works on raw block devices in Linux 7.3, which matters for databases that bypass the filesystem. io_uring also gains dynamic ZCRX areas.

Answer card summarising Ubuntu 26.04.1 LTS: released August 27 2026 as the first point release of the Resolute Raccoon series, bundling four months of stable release updates, with no new hardware enablement kernel until 26.04.2, and opening the automatic upgrade path from Ubuntu 24.04 LTS through Update Manager.

Sysadmin News

Ubuntu 26.04.1 LTS Opens the 24.04 Upgrade Path

Canonical published Ubuntu 26.04.1 LTS on August 27. It carries four months of fixes, no new HWE kernel, and it is the release that starts 24.04 LTS upgrade prompts.

Answer card summarising Asahi Linux M3 progress: an official release with Apple M3 support is close, running on downstream code rather than mainline Linux, with webcams, microphones, USB 3.0, Thunderbolt and hardware video decode working, display support nearly at M1 and M2 parity, and GPU support still outstanding.

Sysadmin News

Asahi Linux Is Close to an M3 Release, GPU Still Out

Webcams, microphones, USB 3.0, Thunderbolt and hardware video decode now work on Apple M3 Macs. Asahi says a release is near, on downstream code, without the GPU.

Answer card summarising the Huawei and HP patent agreement: a multiyear global cross-licensing deal announced August 26 2026 covering Wi-Fi 6 and Wi-Fi 7 patents, with HP paying an undisclosed balancing fee and granting Huawei reciprocal rights, ending a Unified Patent Court case filed in August 2025.

Network News

Huawei and HP Settle Wi-Fi Patents, Wi-Fi 7 Included

HP pays Huawei an undisclosed balancing fee for Wi-Fi 6 and Wi-Fi 7 patents, ending a dispute that started at the Unified Patent Court in August 2025.

Answer card summarising OpenAI site tools: WebMCP support landed in the ChatGPT desktop browser on August 25 2026, letting a page register named tools with document.modelContext and a JSON Schema, available in ChatGPT Work and Codex, with tools inside iframes and HTML form attributes not discovered.

Dev News

OpenAI Puts WebMCP Site Tools in the ChatGPT Browser

OpenAI shipped WebMCP support on August 25. A site can register real tools with document.modelContext instead of hoping an agent clicks the right button.

Answer card summarising the Anthropic and Nscale agreement reported on August 26, 2026: a six year deal worth about 45 billion dollars for roughly 460 megawatts at a West Virginia campus, running Nvidia Vera Rubin systems from late 2027, inside a campus designed for about 1.35 gigawatts across three buildings.

Sysadmin News

Anthropic Rents 460 MW in West Virginia for $45 Billion

Anthropic signed a six year, $45 billion deal for about 460 MW at Nscale's West Virginia campus, on Nvidia Vera Rubin silicon, starting late 2027.

Answer card summarising the AWS and Nvidia announcement of August 26, 2026: two million additional Blackwell Ultra, Rubin and Rubin Ultra GPUs deployed across 2027 and 2028, five months after a commitment of one million GPUs, alongside Vera CPUs, networking and Nemotron open models.

Sysadmin News

AWS Adds Two Million More Nvidia GPUs in 2027 and 2028

AWS will deploy two million more Nvidia Blackwell Ultra, Rubin and Rubin Ultra GPUs across 2027 and 2028, five months after committing to its first million.

Answer card summarising the Linux 7.3 ksmbd changes: opt-in Apple AAPL SMB2 extensions covering Finder compatibility, stream handling, COPYCHUNK, sparse files, CHANGE_NOTIFY and RPC compatibility bits, which together let macOS Time Machine back up to the in-kernel SMB3 server, plus per share SMB3 encryption and SMB3 request replay.

Network News

Linux 7.3 ksmbd Learns to Take Time Machine Backups

Linux 7.3 adds opt-in Apple AAPL extensions to ksmbd, so macOS Time Machine can back up to the in-kernel SMB3 server. The extensions were reverse engineered.

Answer card summarising the in-kernel zstd BMI2 patches posted on August 26, 2026: the kernel re-probed BMI2 CPU support on every compression context creation, three patches probe it once instead, giving a 71 percent reduction in decompression time and an 18 percent reduction in compression time on a crypto_acomp benchmark.

Dev News

Probing BMI2 Once Cuts Kernel Zstd Decode Time by 71%

Three patches stop in-kernel zstd from re-probing BMI2 on every context creation, cutting decompression time by 71% and compression time by 18%.

Answer card summarising the AWS acquisition of DuckLabs announced in late August 2026: AWS buys the Amsterdam company, DuckDB and DuckLake and Quack stay MIT licensed under the independent non profit DuckDB Foundation, founders Hannes Muhleisen and Mark Raasveldt keep leading the project, and closing is expected in early September 2026.

Dev News

AWS Buys the DuckDB Team, MIT License Untouched

AWS has signed to acquire DuckLabs, the Amsterdam team behind DuckDB. The database itself is not part of the deal and stays MIT under the DuckDB Foundation.

Answer card summarising the NFS changes merged for the Linux 7.3 window in late August 2026: NFSD gains CB_NOTIFY support for NFSv4.1 directory delegations driven by fsnotify events covering additions, removals, renames and attribute changes, following the recall only implementation that landed in 2025, alongside continued Netlink work and the removal of the alternative NFSD service thread pool modes.

Network News

Linux 7.3 NFSD Learns to Notify, Not Just Recall

Linux 7.3 adds CB_NOTIFY to NFSD directory delegations, so a changed directory sends the client a description of the change instead of tearing the cache down.

Answer card summarising the reports of August 26 and 27, 2026: The Information says Nvidia agreed to acquire Hugging Face for 12.9 billion dollars, Business Insider says talks valuing the company above 13 billion dollars have not produced a signed agreement, Hugging Face last raised at a 4.5 billion dollar valuation in 2023 with Nvidia among the investors, and the Hub hosts more than three million models.

Dev News

Nvidia Reported to Buy Hugging Face for $12.9 Billion

The Information says Nvidia agreed to acquire Hugging Face for $12.9 billion. Business Insider says nothing is signed yet. Neither company has commented.

Answer card summarising the PorteuX wmbench results published in late August 2026: every benchmark binary carries both a native X11 and a native Wayland backend and chooses at startup, frame rate is measured flat out while CPU and power are measured with the workload locked at 60 frames per second, Wayland led on frame rate, and the lowest power and CPU came from LXDE with Openbox followed by Xfce with the xfwm4-gl fork.

Sysadmin News

Wayland Wins on Frames, X11 Wins on Watts

PorteuX published wmbench, a window manager benchmark with real X11 and Wayland backends. Wayland led on frame rate. The lowest power and CPU were X11 sessions.

Answer card summarising the Cisco Secure AI Factory rack scale expansion announced on August 25, 2026: Supermicro liquid and air cooled compute added, Cisco Nexus One unifying Silicon One front end switches with Nvidia Spectrum-X back end switches, support for Vera Rubin NVL72 and HGX Rubin NVL8, rack to fabric liquid cooling, and orders opening in October 2026.

Network News

Cisco Unifies Two Rival Switch Silicons in One AI Rack

Cisco expanded its Secure AI Factory on August 25, 2026 with Supermicro compute and a Nexus One fabric running Silicon One at the front and Spectrum-X at the back.

Answer card summarising the Emerald AI funding round announced on August 25, 2026: a 150 million dollar Series A at a 1.05 billion dollar valuation, co led by Energize Capital and DCVC, more than 220 million dollars raised in total, and a claim that flexible data centre power draw could unlock over 100 gigawatts on the existing United States grid.

Sysadmin News

Emerald AI Raises $150M to Make Data Centres Flex on Demand

Emerald AI closed a $150 million Series A on August 25, 2026 at a $1.05 billion valuation. Its pitch: flexible power draw could unlock over 100 GW on the US grid.

Answer card summarising MetaRoCE, the RDMA transport Meta published on August 24, 2026: no priority flow control, packets sprayed out of order across multiple planes, receiver driven congestion control in the NIC, about 86 percent throughput retained at 1 percent packet loss and useful bandwidth still flowing at 10 percent loss.

Network News

MetaRoCE Drops PFC and Treats the AI Fabric as Lossy

Meta published MetaRoCE on August 24, 2026. The RDMA transport sprays packets out of order, drops PFC, and holds 86 percent of throughput at 1 percent loss.

Answer card summarising the Nvidia Groq 3 LPX announcement of August 24, 2026: a dedicated inference decode accelerator in full production, 256 accelerators per rack with direct chip to chip links, 3,400 output tokens per second on a 100,000 token context, and up to 35 times the inference throughput per megawatt of Blackwell NVL72.

Sysadmin News

Nvidia Puts Groq 3 LPX Into Full Production for Inference

Announced at Hot Chips on August 24, 2026, Groq 3 LPX enters full production: 256 accelerators per rack, 3,400 output tokens per second on a 100,000 token context.

Answer card summarising the closure of Amazon Mechanical Turk: the service shuts on September 30, 2026 after twenty one years, new customers were refused from July 30, 2026, the marketplace peaked above five hundred thousand workers, and replacements are specialised labelling firms plus models doing their own classification.

Dev News

Amazon Shuts Mechanical Turk on September 30 After 21 Years

Amazon confirmed Mechanical Turk closes on September 30, 2026. If any pipeline of yours still posts HITs, you have about five weeks to move it.

Answer card summarising the IBM Granite 4.2 release of August 25, 2026: dense decoder only models at 3B, 8B and 30B parameters under Apache 2.0, a thinking, non thinking and low effort switch on every model, 128,000 token native context extending to 512,000, and agentic reinforcement learning in real repository, terminal and web search sandboxes for the 8B and 30B.

Dev News

IBM's Granite 4.2 Puts a Thinking Switch in a 3B Model

IBM released Granite 4.2 on August 25, 2026: dense 3B, 8B and 30B reasoning models under Apache 2.0, with a thinking toggle and agentic RL in real sandboxes.

Answer card summarising the OpenAI Jalapeño benchmarks published on August 25, 2026: a first generation inference only accelerator designed with Broadcom on HBM4 memory, 1.5x to 1.9x more AI work per watt at peak throughput, 1.7x to 3.6x lower end to end latency, and still at engineering sample stage.

Sysadmin News

OpenAI's Jalapeño Beats Blackwell on AI Work Per Watt

OpenAI's first custom inference chip, built with Broadcom, posts 1.5x to 1.9x more AI work per watt than the best shipping systems, in benchmarks published August 25.

Answer card summarising Samsung LPDDR5X-PIM at Hot Chips 2026: sixteen processing in memory blocks inside the DRAM banks, 614 gigabytes per second of internal bandwidth against 76.8 on the conventional side, 2.4 TOPS peak at four bit integer weights, and a Llama 3.1 8B run finishing in 5.4 seconds instead of 12.3 with output at 81.3 tokens per second instead of 27.

Sysadmin News

Samsung Puts MAC Units Inside LPDDR5X and Gets 614 GB/s

At Hot Chips 2026 Samsung detailed LPDDR5X-PIM: compute blocks inside the DRAM banks, 614 GB/s against 76.8 conventionally, and 3.01x token throughput on Llama 3.1 8B.

Official Apple product image of the M5 Max and M5 Ultra chip packages announced on August 25, 2026, showing the quad die UltraFusion layout of M5 Ultra alongside the smaller M5 Max package.

Sysadmin News

Apple's M5 Ultra Puts 512GB of Unified Memory Under a Desk

Apple announced M6 and a quad die M5 Ultra on August 25, 2026. The headline number for anyone running models locally is 512GB of unified memory at 1.2TB/s.

Answer card summarising the ath11k queueing patch series posted on August 24, 2026: nine patches, just over 150 lines, moving the transmit queue back under mac80211 control, with latency under load falling from 155 milliseconds to 22 milliseconds on an IPQ8074 access point.

Network News

Nine Patches Cut ath11k WiFi Latency by Seven Times

A patch series posted on August 24, 2026 hands ath11k queueing back to mac80211. On an IPQ8074 access point, latency under load falls from 155 ms to 22 ms.

Answer card summarising Infineon's acquisition of C2i Semiconductors announced on August 24, 2026: a Bangalore company building software defined multiphase controllers and smart power stages, with terms undisclosed and closing expected in the third quarter of calendar 2026.

Sysadmin News

Infineon Buys C2i Semiconductors for AI Rack Power Delivery

Infineon announced on August 24, 2026 that it is acquiring C2i Semiconductors of Bangalore, whose multiphase controllers sit in the last metres of the power chain.

Answer card summarising the Intel Xe driver fix Linus Torvalds merged on August 21, 2026: a rounding direction error in the flat CCS offset calculation that handed compression metadata to the VRAM allocator, found after 24 debugging patches and 18 kernel boots on a Battlemage G21 card with 16 GiB of memory.

Dev News

Torvalds Traced an Xe Black Screen to One round_up Call

24 debug patches and 18 kernel boots to change round_up to round_down in Intel's Xe driver. Torvalds called it a debug session from hell, helped by an AI.

Answer card: the EFI updates merged for Linux 7.3 enforce a 120 second timeout on EFI runtime service calls, after which the firmware is declared wedged, so a single stuck call no longer holds the EFI runtime lock and blocks efivarfs, NVRAM writes and ACPI handlers until the machine is rebooted.

Sysadmin News

Linux 7.3 Gives Wedged EFI Firmware a 120 Second Limit

Linux 7.3 enforces a 120 second timeout on EFI runtime service calls, so buggy firmware can no longer hang efivarfs, NVRAM writes and ACPI handlers until reboot.

Answer card: the NVDIMM and DAX patches merged for Linux 7.3 are mostly preparation for Micron's FAMFS file system for fabric attached memory, which has been through 13 rounds of patch review, was not submitted for the 7.3 merge window, targets upstream acceptance in 2026, and is built for byte level file based access to shared memory pools around 100 TB and larger on CXL servers.

Sysadmin News

Linux 7.3 Clears the DAX Path for Micron's FAMFS

The NVDIMM and DAX work in Linux 7.3 is groundwork for FAMFS, Micron's file system for fabric attached memory, now 13 review rounds deep and aiming for upstream in 2026.

Answer card: ReactOS merged initial Job Objects support on August 22, 2026 through pull request 7500, adding the Windows kernel primitive that manages groups of processes as a single unit for resource limits, collective termination and security restrictions, with the implementation targeting pre Windows 8 functionality and some features still unimplemented, from code originally written nearly a decade earlier.

Sysadmin News

ReactOS Merges Job Objects, Nearly a Decade Later

ReactOS merged initial Job Objects support through pull request 7500, adding the kernel primitive that browser sandboxes and Windows services rely on to group processes.

Answer card: Alibaba launched the Wan3.0 video generation model on August 24, 2026, producing thirty second videos from documents, spreadsheets, slides and web pages, one day after pricing a placement of 710 million shares at HK$112.70 that raised HK$80 billion or about ten billion US dollars, while the newest open weight Wan release on Hugging Face remains Wan 2.2 under Apache 2.0.

Dev News

Alibaba Ships Wan3.0 Closed After a $10B Share Sale

Alibaba launched its Wan3.0 video model on August 24 after a HK$80 billion placement. Wan 2.2 is still the newest open weight release on the Hub.

Answer card: Business Insider reported on August 23, 2026 that Hugging Face is working with a bank to gauge buyer interest in a sale valuing it at thirteen billion dollars or more, roughly triple its four and a half billion dollar valuation from a 2023 funding round, with no deal agreed and no bidders named, following Stripe's agreement to acquire OpenRouter for over seven billion dollars.

Dev News

Hugging Face Explores a Sale at $13 Billion or More

Hugging Face is working with a bank to gauge buyer interest at $13 billion or more, up from $4.5 billion in 2023, weeks after Stripe agreed to buy OpenRouter.

Official IBM chip layout image of the next generation dual architecture processor for IBM Z and LinuxONE, showing the repeated core blocks and cache arrays of a 2 nanometer design with 11 high performance cores.

Sysadmin News

IBM Puts Arm Instructions Inside the Mainframe Core

At Hot Chips 2026, IBM announced a 2 nm mainframe processor with 11 cores above 5.7 GHz where every core executes both Arm and IBM Z instructions natively.

Answer card: the Omacom Foundation launched on August 21, 2026 with eight million dollars from eight patrons at one million each, plus two million more from two later contributors, to hold Omarchy's trademarks and fund its infrastructure, and became the exclusive sponsor of the Hyprland compositor for three years with a two year extension option, which ends the paid Hyprperks subscription tier and makes its features free.

Sysadmin News

Omacom Foundation Opens With $8M and Funds Hyprland

The nonprofit behind Omarchy launched with $8 million from eight patrons and became the exclusive sponsor of Hyprland, ending the paid Hyprperks tier.

Answer card: AMD published RDNA 4m firmware to linux-firmware.git on August 22 2026, covering GFX 11.7.0 and 11.7.1, PSP 15.0.0 and 15.0.9, SDMA 6.1.4 and VPE 2.0.0, with DCN 4.2 display support, ahead of the Zen 6 Ryzen 500 Medusa Point APUs, and Linux 7.3 with Mesa 26.3 as the working baseline.

Sysadmin News

AMD Lands RDNA 4m Firmware in linux-firmware Before Launch

AMD pushed GFX 11.7 firmware to linux-firmware.git, along with PSP 15, SDMA 6.1.4 and VPE 2.0, months ahead of the Zen 6 Medusa Point APUs that need it.

Answer card: Bloomberg reported on August 21 2026 that Anthropic hired Amir Salek, who started Google's custom silicon effort in 2013 and shipped the first seven TPU generations, to join its compute team reporting to James Bradbury, roughly two weeks after Anthropic confirmed an in-house chip design team with roles advertised between 320,000 and 485,000 dollars.

Dev News

Anthropic Hires the Engineer Who Started Google's TPU

Anthropic has hired Amir Salek, who founded Google's TPU program and shipped its first seven generations, two weeks after confirming an in-house chip design team.

Answer card: the KDE weekly Plasma report of August 22 2026 says the panel auto-hide delay drops from 500 milliseconds to 50 milliseconds in Plasma 6.8, adaptive brightness now learns from manual changes, the lock screen gains selectable authentication methods, System Settings search matches English keywords in any locale, and KWin blocks unusably low resolutions while reporting GPU resets on every GPU.

Sysadmin News

Plasma 6.8 Cuts the Panel Auto-Hide Delay to 50 ms

KDE has shortened the Plasma panel auto-hide delay from 500 ms to 50 ms, added learning to adaptive brightness, and made System Settings search work in English too.

Answer card: Nvidia told server assemblers on August 22 2026 that AI server prices rise more than 15% for systems shipping in early 2027, driven by DRAM and high bandwidth memory costs, with Vera Rubin and Grace Blackwell configurations affected and Microsoft, Google and Oracle among the operators notified.

Sysadmin News

Nvidia Warns of 15% AI Server Price Rises for Early 2027

Nvidia has told server builders that AI system prices climb more than 15% in early 2027 as DRAM and HBM costs surge, affecting Vera Rubin and Grace Blackwell racks.

Answer card: OpenAI cut GPT-5.6 Sol API pricing on August 21 2026, with input falling from $5 to $4 per million tokens, cached input from $0.50 to $0.40, and output from $30 to $20, a promotional rate running through November 21 2026 on the API and on eligible ChatGPT Work and Codex credit plans.

Dev News

OpenAI Cuts GPT-5.6 Sol to $4 and $20 Per Million

OpenAI dropped GPT-5.6 Sol API pricing to $4 per million input tokens and $20 output on August 21, a promotional rate that runs to November 21, 2026.

Answer card: Apprise 1.13 released on August 20 2026 is the final version of the 1.x series, adding Pinglet, Trigv and Pingram support plus AES-GCM encryption for Bark, with maintainers advising projects to pin apprise to at least 1.0.0 and below 2.0.0 before the breaking 2.x release.

Sysadmin News

Apprise 1.13 Ends the 1.x Line, So Pin Your Version

Apprise 1.13 is the last release of the 1.x series. Pin apprise below 2.0.0 before the breaking release lands, and read what changed for Bark, Telegram and Matrix.

Answer card: GitHub's August 17 2026 outage lasted 7 hours and 47 minutes, caused by an Istio sidecar proxy reaching its concurrent processing limit in the Central US data centre while autoscaling monitored only the primary container, with a Visual Studio Code retry loop pushing Copilot token traffic from 9,000 to 100,000 requests per second.

Sysadmin News

GitHub Blames 7h47m Outage on an Istio Sidecar Limit

GitHub's CTO published the August 17 postmortem: an Istio sidecar hit its concurrency ceiling, autoscaling never saw it, and a VS Code retry loop did the rest.

Answer card: NetworkManager 1.58.1 released on August 21 2026 fixes a regression where profiles using key-mgmt=wpa-psk failed to reconnect automatically to WPA3 SAE networks, and adds IPv4 forwarding on modem interfaces, systemd-resolved nameserver filtering, a DHCP option 3 warning and stricter certificate paths for private 802.1X profiles.

Network News

NetworkManager 1.58.1 Fixes a WPA3 Reconnect Bug

NetworkManager 1.58.1 repairs a regression that stopped wpa-psk profiles reconnecting to WPA3 SAE networks, plus DNS, modem and 802.1X profile fixes.

Answer card: Nvidia pays Poolside 6 billion dollars for a non exclusive licence to its Model Factory pipeline, invests 1 billion dollars at a 12 billion dollar pre money valuation and hires 109 employees, while Poolside's three founders stay and the company keeps the right to license the same technology elsewhere.

Dev News

Nvidia Pays Poolside $6B to License Its Model Factory

Nvidia licenses Poolside's model building pipeline for $6 billion, invests $1 billion at a $12 billion valuation, and hires 109 staff. Poolside stays independent.

Answer card: on August 21 2026 AMD maintainer Alex Deucher posted a 95 patch series of about seven thousand lines adding Ultra Accelerator Link infrastructure and AMDGPU integration to the Linux kernel, plus a three patch series exposing the IOCTL interface, targeting scale up pods of up to 1,024 accelerators.

Network News

AMD Posts 95 Patches to Put UALink in the Linux Kernel

Alex Deucher sent roughly 7,000 lines of UALink infrastructure to the kernel list on August 21, aiming at pods of up to 1,024 accelerators.

Answer card: Google announced on August 20 2026 that its Gemma family of open weight models has passed one billion cumulative downloads, with more than one hundred thousand community published variants over two years, a total that excludes Gemma running inside Android and Chrome.

Dev News

Google's Gemma Open Models Pass One Billion Downloads

Gemma crossed a billion downloads on August 20 with over 100,000 community variants, and the count leaves out Android and Chrome entirely.

Answer card: OpenZFS 2.4.4 was released on August 21 2026 with support for the Linux 7.2 kernel taking the supported range to 4.18 through 7.2, a new zhack mmp reclaim command for pools stranded by Multi Modifier Protection, an optional systemd-udev-settle dependency for import units, and fixes for an infinite loop in zfs_write and mmap reads past end of file.

Sysadmin News

OpenZFS 2.4.4 Adds a Way Out of an MMP Stranded Pool

OpenZFS 2.4.4 shipped on August 21 with Linux 7.2 support and a new zhack mmp reclaim command for pools that multihost protection will not release.

Answer card: Wine 11.16 was released on August 21 2026 with a VA-API backend for Direct3D hardware video decoding that currently supports H.264 only, an updated Wine Mono engine with ARM64 support, improved ARM64EC exception handling, and 35 bug fixes including the Steam client and a WineWayland regression.

Sysadmin News

Wine 11.16 Adds VA-API Video Decoding, H.264 Only

Wine 11.16 landed on August 21 with a VA-API backend for Direct3D video decode, Wine Mono on ARM64, and 35 bug fixes including the Steam client.

Answer card: Microsoft will retire the Azure VMware Solution license included SKUs on August 30 2027, license included pay as you go nodes run unchanged only through October 31 2026, and customers must bring their own VMware Cloud Foundation subscription purchased from Broadcom into the AVS VCF BYOL option, available in all 39 AVS regions at a lower price than the bundled SKU.

Sysadmin News

Azure Retires License Included VMware Nodes in 2027

Microsoft confirmed the Azure VMware Solution license included SKUs retire on August 30, 2027. PayGo nodes run unchanged only through October 31, 2026.

Answer card: on August 21 2026 Google made Antigravity available as an extension for Visual Studio Code, Visual Studio, JetBrains IDEs from version 2026.2.1 and Zed, and added it to Gemini Enterprise Standard, Plus and Standard Emerging Market subscriptions with project level budget caps, pooled token quotas, optional overages, audit logging and agent policy controls.

Dev News

Antigravity Ships Extensions for VS Code and JetBrains

Google put Antigravity into VS Code, Visual Studio, JetBrains and Zed, and folded it into Gemini Enterprise with budget caps, pooled tokens and audit logs.

Answer card: the Linux 7.3 networking merge window closed on August 20 2026 with 632 net patches and 648 net-next patches, a third to a half of the net-next set described by maintainer Jakub Kicinski as AI driven low priority work, with Meta funding multi model LLM review to keep up.

Network News

Netdev Is Overwhelmed and Now Reviews Patches With LLMs

Jakub Kicinski merged 1,280 networking patches for Linux 7.3 and says netdev is completely overwhelmed by AI output. Meta paid for the LLM review budget.

Answer card: Rust 1.98 was released on August 20 2026 with algebraic addition, subtraction, multiplication, division and remainder methods on f32 and f64 that allow fast math style reordering and vectorisation per operation, plus a format_into method using NumBuffer for allocation free integer formatting.

Dev News

Rust 1.98 Adds Algebraic Float Math to Close an 8x Gap

Rust 1.98 landed on August 20 with algebraic methods on f32 and f64, opt in fast math that unlocks vectorisation, plus buffered integer formatting.

Answer card: Btrfs updates merged for Linux 7.3 on August 19, 2026 raise direct I/O from about 50 percent to about 95 percent of theoretical maximum using an iomap bounce buffer, give about 3x from replacing XArray with a local LRU list, about 5x on logging throughput by removing a one jiffy delay, and about 5x on full fsync for files without holes.

Sysadmin News

Btrfs in Linux 7.3 Doubles Direct I/O and Speeds fsync 5x

David Sterba sent the Btrfs updates for Linux 7.3 on August 19, 2026. Direct I/O climbs from about 50 to about 95 percent of theoretical maximum.

Answer card: Peter Zijlstra's cgroup scheduling rework merged for Linux 7.3 in August 2026 moves EEVDF to a single runqueue in certain cgroup situations, adds four weight distribution modes named up, max, concur and tasks with concur as the default, and lifted minimum frame rate from 4 to 29 fps on an Intel Core i7-2600K with a Radeon RX 580.

Sysadmin News

Linux 7.3 Rebuilds cgroup Scheduling on One Runqueue

Peter Zijlstra's cgroup scheduling rework merged for Linux 7.3 moves EEVDF to a single runqueue and adds four weight distribution modes, with concur as the default.

Answer card: Marvell issued Google a warrant on August 18, 2026 for 58,970,907 shares at 206.58 dollars, about 12.2 billion dollars if fully exercised, vesting one tranche per 500 million dollars of custom chip purchases across 240 tranches from Marvell's third quarter of fiscal 2027 through the end of fiscal 2033.

Sysadmin News

Marvell Hands Google a $12.2B Warrant for TPU Silicon

An SEC filing shows Marvell issued Google a warrant for 58,970,907 shares at $206.58, vesting one tranche per $500M of custom chip purchases through fiscal 2033.

Answer card: Veeda AI announced a seed round of more than 90 million dollars on August 19, 2026, co-led by Khosla Ventures and Radical Ventures, three months after being incorporated in early June 2026, founded by former Nvidia researchers Sanja Fidler as chief executive, Zan Gojcic as chief technology officer and Huan Ling as chief scientist, to build multimodal world models for training embodied AI agents.

Dev News

Veeda AI Raises $90M Seed to Build World Models

Sanja Fidler, who led Nvidia's Spatial Intelligence Lab, raised more than $90 million co-led by Khosla and Radical, three months after founding Veeda in Toronto.

Answer card: OpenAI begins serving ChatGPT Ads in 31 European countries on August 24, 2026, limited to the Free and Go plans while Plus, Pro and Enterprise remain ad free, with ad selection at launch drawing only on the current conversation topic, approximate location, device type, time of day and language, and with past chats and stored memories explicitly excluded.

Dev News

ChatGPT Ads Reach 31 European Markets on August 24

OpenAI starts serving ads in ChatGPT across 31 European countries on August 24, 2026, on the Free and Go plans only, with personalization off by default at launch.

Answer card: Go 1.27 was released on August 19, 2026 with generic methods, struct literals accepting any valid field selector, a new encoding/json/v2 engine now backing the existing encoding/json package, a standard library uuid package, and size specialized allocation cutting the cost of allocations under 80 bytes by about 30 percent.

Dev News

Go 1.27 Puts json/v2 Under the Old encoding/json API

Go 1.27 arrived on August 19, 2026 with generic methods, a stdlib uuid package and encoding/json quietly rebuilt on a v2 engine, with a temporary opt out flag.

Answer card: the Linux 7.3 memory management pull from Andrew Morton reduces locking contention in the zsmalloc zs_free path used by zRAM and Zswap, measured at up to 1.83x on a Raspberry Pi 4B and up to 1.4x on a 20 core Intel system, and rewrites rmap_walk_ksm so a worst case lock hold falls from 705 milliseconds to 1.67 milliseconds.

Sysadmin News

Linux 7.3 Cuts a 705 ms KSM Stall Down to 1.67 ms

The Linux 7.3 memory management pull cuts zsmalloc lock contention by up to 1.83x on a Raspberry Pi 4B and drops a worst case KSM lock hold from 705 ms to 1.67 ms.

Answer card: Cursor started rolling out Origin, its code hosting platform, to all paid plans on August 17, 2026, with repositories, pull requests, two way GitHub sync, and Vercel, Depot and Buildkite integrations available on day one.

Dev News

Cursor Ships Origin, Its Code Hosting GitHub Rival

Cursor started rolling out Origin, its code hosting platform, to paid plans on August 17, 2026. It syncs with GitHub and runs existing GitHub Actions unchanged.

Answer card: on August 18, 2026 Etched raised 700 million dollars at a 21 billion dollar valuation led by Jane Street, double its July valuation of 10.3 billion dollars, with more than 1 billion dollars in booked orders and its first cluster installed in Jane Street's data center.

Sysadmin News

Etched Raises $700M at $21B and Ships to Jane Street

Etched raised $700 million at a $21 billion valuation on August 18, 2026. Jane Street led the round and installed the first Etched cluster in its own data center.

Answer card: Warp launched Factories on August 18, 2026, a closed beta system defining agent pipelines as version controlled YAML across triage, spec, implementation, review and verification, with bring your own inference and compute, zero data retention options and a cost per pull request dashboard.

Dev News

Warp Factories Puts the Whole SDLC on Agent Rails

Warp launched Factories on August 18, 2026: version controlled YAML pipelines that run triage, spec, implementation and review agents on your own inference and compute.

Answer card: Cerebras announced the CS-4 on August 18, 2026 with three WSE-3 Turbo wafers in one rack, 750 petaflops of AI compute, 129.6 petabytes per second of aggregate memory bandwidth, 7.2 terabits per second of system I/O and wafer to wafer latency as low as 2 microseconds, shipping in the third quarter of 2026.

Sysadmin News

Cerebras CS-4 Puts Three Wafers in One Rack at 750 PFLOPS

Cerebras announced the CS-4 on August 18, 2026. Three WSE-3 Turbo wafers in one rack, 750 PFLOPS, 129.6 PB/s of memory bandwidth and RoCE v2 networking.

Answer card: a JLL report covered by Reuters on August 19, 2026 finds European data centre projects planned for 2026 to 2028 average 175 kilometres from major hub cities against 46 kilometres for campuses delivered between 2022 and 2025, with greenfield sites rising to 39 percent of the pipeline from 8 percent and inner city projects falling to 5 percent from 13 percent.

Network News

Europe's AI Data Centres Move 175 km Away From the Hubs

A JLL report reported by Reuters on August 19, 2026 finds European AI data centres planned for 2026 to 2028 sit 175 km from major hubs, against 46 km before.

Answer card: the Financial Times reported on August 19, 2026 that ByteDance and Tencent each received about 10,000 Nvidia H200 accelerators in recent weeks, against United States clearance for up to 100,000 each, with every purchase requiring case by case approval from China's National Development and Reform Commission and most licensed orders steered to Hong Kong where neither company runs data centres.

Sysadmin News

First Nvidia H200s Reach China, 10,000 Each to Two Buyers

The FT reported on August 19, 2026 that ByteDance and Tencent each took about 10,000 Nvidia H200 GPUs, with Beijing steering most licensed orders to Hong Kong.

Answer card: Reuters reported on August 18, 2026 that Samsung raised foundry prices in July, with the 4 nanometre SF4 process up 10 to 15 percent for US and China customers and 5 to 10 percent for Taiwan customers, the 5 nanometre SF5 process up 10 to 15 percent, and the 8 nanometre node up nearly 10 percent, while its Pyeongtaek SF4 line has run at full capacity since late 2025.

Sysadmin News

Samsung Raises Foundry Prices by Up to 15% on AI Demand

Reuters reported on August 18, 2026 that Samsung raised 4nm and 5nm foundry prices by 10 to 15 percent in July, with 8nm up nearly 10 percent, as TSMC fills up.

Answer card: a Wall Street Journal analysis published August 17, 2026 found nine technology companies disclosing about 3 trillion dollars of off balance sheet AI commitments, made up of 1.2 trillion in leases not yet commenced and 1.9 trillion in purchase commitments, against roughly 600 billion dollars of reported capital expenditure over the latest twelve months.

Sysadmin News

Big Tech Holds $3T in AI Commitments Off Balance Sheet

A Wall Street Journal analysis published August 17, 2026 found nine tech firms disclose about $3 trillion of AI commitments in filing footnotes, off balance sheet.

Answer card: Groq closed a 350 million dollar Series A on August 17, 2026 at a 3.5 billion dollar valuation, down from 6.9 billion in September 2025, led by Disruptive with planned participation from Nvidia, funding a build from about 54 megawatts across 13 datacenters toward more than 200 megawatts during 2027.

Sysadmin News

Groq Raises $350M at $3.5B, Half Its 2025 Valuation

Groq closed a $350 million Series A on August 17, 2026 at a $3.5 billion valuation, down from $6.9B, to grow its inference cloud from 54 MW toward 200 MW.

Answer card: Wispr closed a 280 million dollar Series B at a 2 billion dollar valuation on August 17, 2026, led by Menlo Ventures, bringing total funding to 361 million dollars, and shipped the Canto speech model which the company says takes noisy-condition error rates from over 30 percent of words down to between 5 and 10 percent.

Dev News

Wispr Raises $280M and Ships a New Speech Model

Wispr closed a $280 million Series B at a $2 billion valuation on August 17, 2026, and shipped Canto, a speech model it says cuts noisy-room errors to 5-10%.

Answer card: FESCo rejected the Build x86-64-v3 Packages change for Fedora 45 after months of debate, asking for more performance data and a plan for install media, with resubmission possible for Fedora 46, while RHEL 10 already requires x86-64-v3 and Microsoft wants that baseline for the Fedora based Azure Linux.

Sysadmin News

FESCo Rejects x86-64-v3 Packages for Fedora 45

Fedora will not ship x86-64-v3 optimised packages in Fedora 45. FESCo rejected the change and asked for more performance data before reconsidering it for Fedora 46.

Answer card: Firefox 154 shipped on August 18, 2026 with text-box-trim, text-box-edge and the text-box shorthand, sibling-index() and sibling-count(), a native time and datetime-local picker, No-Vary-Search disk cache support, four new iterator methods, and a permission prompt before a page opens a WebSocket to a local network device.

Dev News

Firefox 154: New CSS, Iterator Helpers, Local WebSockets

Firefox 154 ships text-box-trim, sibling-index(), native time pickers and new iterator methods, and now asks permission before a page opens a WebSocket to your LAN.

Answer card: Linux 7.3 merged FailFS, 931 lines including documentation, where every operation returns EOPNOTSUPP and the root cannot be opened even with O_PATH, alongside a new fchroot system call and the FD_FAILFS_ROOT sentinel, while the EFS and FreeVxFS drivers were deleted for about 3700 lines removed.

Sysadmin News

Linux 7.3 Merges FailFS and Drops Two 1990s File Systems

Linux 7.3 merged FailFS, a file system where every operation returns EOPNOTSUPP, plus a new fchroot() syscall, and deleted the EFS and FreeVxFS drivers.

Answer card: the three Imagen 4 endpoints in the Gemini API shut down on August 17, 2026, with gemini-3.1-flash-image as the recommended replacement, and four parameters removed including numberOfImages, negativePrompt, imageFormat and addWatermark.

Dev News

Google Shuts Down Its Imagen 4 API Endpoints

The three Imagen 4 endpoints in the Gemini API stop responding on August 17, 2026. The replacement is not a drop-in, and four parameters disappear entirely.

Answer card: Linux 7.2 was released on August 16, 2026 with cache aware scheduling, the USB4STREAM protocol, faster ext4 and Btrfs paths, MultiPath TCP raised to 64 subflows, and Rust support on the s390 architecture.

Sysadmin News

Linux 7.2 Released With Cache Aware Scheduling

Linus Torvalds shipped Linux 7.2 on August 16, 2026 with cache aware load balancing, USB4STREAM, faster ext4 and Btrfs, and 64 MPTCP subflows.

Answer card: Bloomberg reported on August 16, 2026 that Stripe agreed to acquire OpenRouter for more than 7 billion dollars, over five times the 1.3 billion dollar valuation from its May Series B and below the roughly 10 billion dollars reported in July talks.

Dev News

Stripe Agrees to Buy OpenRouter for Over $7 Billion

Bloomberg reported on August 16, 2026 that Stripe agreed to acquire OpenRouter for more than $7 billion, well below the $10 billion figure floated in July.

Answer card: Higgsfield announced a 400 million dollar Series B on August 17, 2026 at a 5.4 billion dollar valuation led by DST Global, roughly four times its 1.3 billion dollar valuation six months earlier, with 700 million dollars in annualized revenue and over 30 million users in 238 countries.

Dev News

Higgsfield Raises $400M at a $5.4 Billion Valuation

Higgsfield announced a $400 million Series B on August 17, 2026 at a $5.4 billion valuation, with annualized revenue reaching $700 million and 30 million users.

Answer card: the Guardian reported on August 17, 2026 that internal documents put Microsoft at about 2.2 million installed AI chips after roughly 280 billion dollars of spending since 2022, against an internal target of 1.8 million by the end of 2024, with fewer than 500,000 Blackwell parts installed. Microsoft calls the calculations inaccurate.

Sysadmin News

Microsoft Runs 2.2M AI Chips After a $280B Spend

The Guardian reported on August 17, 2026 that internal documents put Microsoft at 2.2 million installed AI chips after $280 billion of spending. Microsoft disputes it.

Answer card: on August 17, 2026 Nvidia confirmed residual value guaranties capped at 105 billion dollars for the PORTS-Pike Technology Campus in Pike County, Ohio, backing an initial 4.25 IT-GW with an option on a further 3.75 IT-GW, built and operated by SB Energy under a 20 year lease to OpenAI, with first capacity in 2028.

Sysadmin News

Nvidia Guarantees $105B for OpenAI's Ohio AI Campus

Nvidia confirmed on August 17, 2026 that it will backstop up to $105 billion of land, power and shell buildout at PORTS-Pike, plus a $1.5B stake in SB Energy.

Answer card: Anthropic reported preliminary second quarter 2026 revenue above eleven and a half billion dollars, against seven hundred and eighty seven million a year earlier and four point seven three billion in the first quarter, with positive adjusted operating income ahead of a possible autumn listing.

Dev News

Anthropic's Quarterly Revenue Passes $11.5 Billion

Preliminary figures put Anthropic at more than $11.5 billion for the second quarter, up from $787 million a year earlier, with positive adjusted operating income.

Answer card: Debian's General Resolution on LLM usage went to a vote from August 15 to August 28, 2026 with eight ballot options from a Social Contract prohibition needing a three to one supermajority to permissive frameworks needing a simple majority.

Dev News

Debian's LLM Vote Opens With Eight Ballot Options

Voting runs from August 15 to August 28, 2026. The ballot grew from four proposals to eight, and only the Social Contract amendment needs a 3:1 majority.

Answer card: Nvidia's 13F filed August 14, 2026 discloses about 122.8 million SpaceX Class A shares worth roughly 21 billion dollars and 214.8 million Intel shares worth about 30 billion from a 5 billion dollar cost basis, together close to 80 percent of the disclosed portfolio, alongside a reduced OpenAI backstop.

Sysadmin News

Nvidia's Filing Shows $21B in SpaceX, $30B in Intel

A 13F filed on August 14 puts two customers at roughly 80% of Nvidia's disclosed equity portfolio, and the OpenAI backstop was cut to under $120 billion.

Answer card: SpaceX closed its sixty billion dollar all stock acquisition of Anysphere, the maker of Cursor, on August 14, 2026, giving the editor access to SpaceX GPU capacity while its staff are distributed across SpaceXAI teams rather than kept as a standalone unit.

Dev News

SpaceX Closes Its $60 Billion Cursor Acquisition

The all stock purchase of Anysphere completed on August 14, 2026. Cursor gains SpaceX GPU capacity, and its team folds into the SpaceXAI organisation.

Answer card describing Intel patches queued for the Linux 7.3 merge window that fix cluster aware load balancing on hybrid CPUs, where remaining tasks fail to spread evenly across E core clusters.

Sysadmin News

Linux 7.3 Set to Fix Balancing on Intel Hybrid CPUs

Intel patches queued for the Linux 7.3 merge window fix cluster load balancing that has been spreading work unevenly across E core clusters since 2021.

Answer card describing the merge of consumer Copilot and Microsoft 365 Copilot into a single app, with Group Chat, Podcasts and consumer Deep Research retired on August 18, 2026 and Group Chat content permanently deleted.

Sysadmin News

Copilot Merges Into One App, Three Features Retire

Microsoft began merging consumer Copilot and Microsoft 365 Copilot into one app, and on August 18, 2026 it retires Group Chat, Podcasts and consumer Deep Research.

Answer card describing the Qwen3.8-27B release of August 14, 2026, a 27 billion parameter dense multimodal model published under Apache 2.0 with 262,144 tokens of native context, extensible to one million with YaRN.

Dev News

Qwen3.8-27B Ships Open Weights Under Apache 2.0

Alibaba released the 27B companion to Qwen3.8-Max on August 14, 2026, a dense multimodal model with 262,144 tokens of native context under Apache 2.0.

Answer card describing Apple's August 13, 2026 court filing proposing commissions of 15% for standard apps, 10% for subscription renewals and partner programs, and 5% for Small Business Program members on purchases made through external links in iOS apps.

Dev News

Apple Proposes 15% on App Store Link Out Purchases

Apple asked a US court on August 13, 2026 to let it charge 15%, 10% or 5% on purchases made through external links in iOS apps. Epic argues the number should be zero.

Answer card explaining that from August 14, 2026 Gemini and Flow users can switch off the visible AI watermark on images, videos and songs, while the invisible SynthID signal and C2PA provenance metadata remain embedded in every generation.

Dev News

Gemini Adds a Toggle to Hide Visible AI Watermarks

Google let Gemini and Flow users switch off the visible sparkle on AI images, video and music from August 14, 2026. SynthID and C2PA metadata stay in every file.

Answer card describing the RustDesk preview build of August 14, 2026 that adds true unattended remote access on Wayland with multi monitor support, shipped for x86_64 Debian and Ubuntu based systems.

Sysadmin News

RustDesk Brings Unattended Remote Access to Wayland

RustDesk published a preview build on August 14, 2026 that connects to a Wayland machine with nobody there to approve the session, multi monitor included.

Answer card describing Blacksmith's $45 million Series B announced on August 12, 2026, led by Peak XV Partners at a $550 million valuation, with more than 6,000 customers using its GitHub Actions runners, up from about 800 at its Series A in September 2025, and claims of twice the speed at 60 percent less cost than GitHub hosted runners.

Dev News

Blacksmith Raises $45M as CI Becomes the New Bottleneck

Blacksmith closed a $45 million Series B at a $550 million valuation. Its GitHub Actions runners now serve more than 6,000 companies, up from about 800.

Answer card describing the ninth generation 2Tb QLC 3D flash memory device unveiled by Kioxia and Sandisk on August 12, 2026, with four bits per cell, a six plane architecture, a 4.8 Gb/s NAND interface that is 33 percent faster than the eighth generation, and CMOS directly Bonded to Array construction.

Sysadmin News

Kioxia and Sandisk Show a 2Tb QLC NAND Running at 4.8 Gb/s

Kioxia and Sandisk unveiled a ninth generation 2Tb QLC 3D flash device on August 12, 2026, with a six plane design and a 4.8 Gb/s interface, 33 percent faster.

Answer card describing the Ultrafast service tier previewed by OpenAI on August 13, 2026, running GPT-5.6 Sol on Cerebras wafer scale hardware at up to 750 output tokens per second, up to 14 times faster than Standard processing, in limited preview in the OpenAI API.

Dev News

OpenAI Previews Ultrafast, GPT-5.6 Sol at 14x Speed

OpenAI previewed Ultrafast on August 13, 2026, a new API service tier that runs GPT-5.6 Sol on Cerebras hardware at up to 750 output tokens a second.

Answer card describing the Podman 6.1 release of August 13, 2026, adding podman volume rename, podman machine restart, a network rm --ignore flag, manifest push retry options, a Quadlet ImageVolume key and a fix for a race condition that could corrupt generated systemd units.

Sysadmin News

Podman 6.1 Adds Volume Rename and Fixes a Quadlet Race

Podman 6.1 landed on August 13, 2026 with volume rename, machine restart, an ImageVolume key for Quadlet, and a fix for a race that corrupted systemd units.

Answer card summarising the Databricks round of August 13, 2026: $5 billion raised at a $190 billion valuation led by Coatue, a revenue run rate above $7 billion growing more than 80 percent year over year, and more than 1,000 customers past $1 million in annual recurring revenue.

Dev News

Databricks Raises $5B at $190B, Aiming It All at Agents

Databricks closed a $5 billion round at a $190 billion valuation on August 13, 2026, crossing a $7 billion revenue run rate with more than 80 percent growth.

Answer card describing the DeepSeek API repricing effective August 16, 2026, with V4-Pro output moving from $0.87 per million tokens to $1.98 off peak and $3.96 at peak, peak hours running 01:00 to 04:00 and 06:00 to 10:00 UTC, and cache hit prices rising by as much as 1,100 percent.

Dev News

DeepSeek V4 Pro Raises API Prices and Adds Peak Hours

DeepSeek shipped V4-Pro-0813 and repriced its API from August 16, 2026, with peak and off peak rates and cache hit costs rising by as much as 1,100 percent.

Answer card summarising Gemini 3.7 Flash, released by Google on August 13, 2026 at an introductory price of $0.75 per million input tokens and $3.75 per million output tokens, with a 1 million token context window and a jump on DeepSWE v1.1 from 49.0 to 65.3 percent.

Dev News

Gemini 3.7 Flash Ships Three Weeks After 3.6, Half Price

Google shipped Gemini 3.7 Flash on August 13, 2026 at $0.75 and $3.75 per million tokens, an introductory rate that doubles on January 1, 2027.

Answer card describing the Sysinternals 2026.8 refresh of August 12, 2026, covering seven tools including RDCMan 3.20, which saves RDG files that older program versions cannot open and backs up legacy files as filename.old.

Sysadmin News

Sysinternals 2026.8: RDCMan 3.20 Rewrites Your RDG Files

Microsoft updated seven Sysinternals tools on August 12, 2026. RDCMan 3.20 saves RDG files that older builds cannot open, which matters for shared server lists.

Answer card describing the Comma.ai Tiny Chestnut dock announced on August 12, 2026, a PCIe Gen4 x4 to USB4 enclosure at 249 dollars or 799 dollars with a Radeon RX 9060, built on the ASMedia ASM2464PD controller and running open source C firmware published in the tinygrad GitHub organisation.

Sysadmin News

Comma.ai Ships a $249 USB4 eGPU Dock With Open Firmware

Comma.ai launched the Tiny Chestnut on August 12, 2026: a PCIe Gen4 x4 to USB4 dock at $249, running open source C firmware for the ASM2464PD controller.

Answer card describing Google's announcement of August 11, 2026 that the Gemini app passed 1 billion monthly active users, its fourteenth product to reach that scale, with 63 percent of users interacting by voice, one in five Gemini Live sessions using live camera or screen sharing, and more than 150 million images generated per day.

Dev News

Gemini Hits 1 Billion Monthly Users, and 63% Talk to It

Google says the Gemini app passed 1 billion monthly active users, its 14th product to do so. The usage split matters more than the milestone: 63% use voice.

Answer card describing the IBM and Together AI agreement signed on August 11, 2026, a 240 million dollar multi-year deal for an inference cluster on IBM Cloud built from around 2000 Nvidia HGX B300 Blackwell systems connected with Spectrum-X Ethernet, aimed at serving open weight models and expected to be available in the first quarter of 2027.

Sysadmin News

IBM Puts $240M Into a Together AI Inference Cluster

IBM and Together AI signed a $240 million multi-year deal for an inference cluster on IBM Cloud, built on Nvidia HGX B300 systems with Spectrum-X Ethernet.

Answer card describing the Lovable Series C announced on August 12, 2026, a 400 million dollar round at a 13.3 billion dollar valuation led by Menlo Ventures and co-led by the Scaleup Europe Fund, with 60 million projects built since November 2024 and around 900 million monthly visits to apps made on the platform.

Dev News

Lovable Raises $400M at a $13.3 Billion Valuation

Lovable closed a $400 million Series C on August 12, 2026 at a $13.3 billion valuation, double December, with Menlo Ventures and the EU backed Scaleup Europe Fund.

Answer card describing Anthropic's confirmation on August 11, 2026 that every Claude model launched on or after August 2, 2026 embeds a statistical watermark in generated text across the platform API, claude.ai, Claude Code, Claude Cowork and Claude Tag, with C2PA signed manifests on supported files and no opt out available to users or developers.

Dev News

Claude Now Watermarks Every Output, With No Opt Out

Anthropic embeds a statistical watermark in text from every Claude model launched since August 2, 2026. There is no API flag, plan tier or setting to disable it.

Answer card describing CodeRabbit's $143 million Series C announced on August 12, 2026 at a $1.5 billion valuation, co-led by Atomico and Smash Capital, with more than 2 million code reviews performed each week, over 17,000 customers, 150,000 open source projects and revenue growth of more than fivefold year over year.

Dev News

CodeRabbit Raises $143M at $1.5B, 2M Reviews a Week

CodeRabbit closed a $143 million Series C at a $1.5 billion valuation. It now runs more than 2 million code reviews a week for over 17,000 customers.

Answer card describing the Manjaro 26.1 Bian-May release of August 12, 2026, the first feature release since January 2026, shipping GNOME 50, KDE Plasma 6.7 and Xfce 4.20 on Linux 7.1 with Linux 6.18 LTS and 6.12 LTS available as alternative kernels.

Sysadmin News

Manjaro 26.1 Ships GNOME 50, Plasma 6.7 and Linux 7.1

Manjaro 26.1 Bian-May landed on August 12, 2026 with GNOME 50, KDE Plasma 6.7, Xfce 4.20 and Linux 7.1 by default, plus 6.18 and 6.12 LTS kernels.

Answer card describing Rik van Riel's request for comments patch series that batches lookups in the Linux kernel follow_page_mask function so page table walkers can return more than one page per call, benefiting get_user_pages callers, with the kernel gup_test benchmark reporting improvements of up to 12.8 times.

Sysadmin News

A Linux MM Patch Set Shows Up to 12.8x Gains in gup_test

Meta engineer Rik van Riel has RFC patches letting follow_page_mask return more than one page per call. The kernel gup_test benchmark shows up to 12.8x improvement.

Answer card describing Nvidia's announcement of August 11, 2026 that it signed memorandums of understanding with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to mobilise more than 500 billion dollars of third party capital for AI datacenter buildout, with debt collateralised by compute and Nvidia contributing up to 25 percent of an opportunity.

Sysadmin News

Nvidia Lines Up 500 Billion Dollars to Finance AI Buildout

Nvidia signed memorandums of understanding with six asset managers on August 11, 2026 to mobilise 500 billion dollars for datacenters, with GPUs as collateral.

Answer card describing the ChatGPT desktop app for Linux released in preview on August 11, 2026, shipping native deb and rpm packages for x64 and ARM64, validated on Ubuntu 24.04 and 26.04 LTS, Debian 13 and Fedora 43 and 44, bundling ChatGPT, ChatGPT Work and Codex in one closed source application.

Dev News

OpenAI Ships a ChatGPT Desktop App for Linux, With Codex

OpenAI released the ChatGPT desktop app for Linux in preview on August 11, 2026: deb and rpm packages, x64 and ARM64, with Codex running against local repositories.

Answer card describing QEMU 11.1, released on August 11, 2026 with more than 3200 commits from 285 authors, adding UFS 4.1 Write Booster and Host Initiated Defragmentation emulation, big endian RISC-V support, nested virtualization with vGIC on the Apple hvf accelerator, and vhost-host-user offload for virtio-rtc.

Sysadmin News

QEMU 11.1 Adds UFS Write Booster and Big Endian RISC-V

QEMU 11.1 shipped on August 11, 2026 with over 3200 commits from 285 authors: UFS 4.1 emulation, big endian RISC-V, and nested virtualization on Apple hvf.

Answer card: a three patch Linux series posted on August 12, 2026 adds the missing power lifecycle for Apple GMUX hybrid graphics on the 2018 and 2019 MacBookPro15,1, spanning apple-gmux, amdgpu and the HDA controller, allowing the discrete GPU to transition between DynOff and DynPwr states and cutting idle power draw from about twenty four watts to about twelve.

Sysadmin News

Linux Patches Bring Hybrid Graphics to 2018 MacBook Pros

A three patch series gives the 2018 and 2019 MacBook Pro 15,1 working hybrid graphics on Linux, cutting roughly twelve watts from idle power draw.

Answer card: LACT 0.10 was released on August 12, 2026, adding NVIDIA PowerMizer performance level forcing and a voltage boost slider, a Blackwell specific hotspot sensor path that replaces bogus readings, per memory chip temperature sensors for GDDR6X and GDDR7 including both sides of clamshell cards, a GTT usage bar on AMD, an interactive service setup wizard and polkit authentication for profile hook commands.

Sysadmin News

LACT 0.10 Adds NVIDIA PowerMizer and Blackwell Hotspot

LACT 0.10 adds NVIDIA PowerMizer and voltage boost controls, a working Blackwell hotspot sensor, per memory chip temperatures and a real setup wizard.

Answer card: Linux 7.2 reverts the DRM scheduler default from the fair policy back to FIFO, in a nineteen patch series sent by Tvrtko Ursulin on August 11, 2026, after a regression report described a Radeon RX 9070 XT dropping to roughly ten frames per second under sustained load, with the fair policy now marked experimental.

Sysadmin News

Linux 7.2 Reverts the Fair DRM Scheduler Back to FIFO

Linux 7.2 will not ship with the fair DRM scheduler after all. A nineteen patch revert restores FIFO following an RX 9070 XT regression report.

Answer card: mold 2.42 was released on August 12, 2026, four months after 2.41, adding SFrame version 3 stack unwinding support, Android packed relocation formats APS2 and relr, compression levels on debug sections, the minus w option, default discarding of temporary local symbols, and fixes for three identical code folding correctness bugs.

Dev News

Mold 2.42 Adds SFrame v3 and Android Packed Relocations

Mold 2.42 lands after four quiet months with SFrame v3 unwinding, Android packed relocations, tunable debug compression and three real ICF correctness fixes.

Answer card: the Fedora Engineering and Steering Committee approved ticket 3661 to enable systemd-oomd.service and swap on zram by default in Fedora CoreOS starting with Fedora Linux 45, aligning CoreOS with other Fedora variants, with the change affecting existing running nodes as well as new installs and requiring attention for Kubernetes clusters.

Sysadmin News

Fedora CoreOS Turns On systemd-oomd and zRAM Swap

FESCo approved enabling systemd-oomd and swap on zram by default in Fedora CoreOS for Fedora 45. The change reaches existing running nodes, not only new installs.

Answer card: the Rust DRM pull for the Linux 7.3 merge window brings the Nova driver for NVIDIA GPUs a consolidated GPU System Processor boot process, vGPU boot support, TLV firmware image format handling, and improvements to the Hopper and Blackwell support code, alongside continued work on the Tyr driver for Arm Mali.

Sysadmin News

Nova Gains vGPU Boot and TLV Firmware for Linux 7.3

The Rust based Nova driver for NVIDIA GPUs picks up vGPU boot support, TLV firmware handling and a consolidated GSP boot path in the Linux 7.3 merge window.

Answer card: OpenAI completed a tender offer of roughly 7 billion dollars in employee shares at an 852 billion dollar valuation, unchanged from the March 2026 round, funding the buyback from its own balance sheet rather than selling to outside investors, following a confidential SEC filing for a possible IPO in June 2026.

Dev News

OpenAI Buys Back 7 Billion Dollars of Employee Shares

OpenAI completed a 7 billion dollar employee tender offer at an 852 billion dollar valuation, funding the buyback itself instead of adding outside investors.

Answer card: from February 1, 2027 the YouTube Partner Program requires new applicants to have 1,000 subscribers plus either 8,000 qualified public watch hours over the last 365 days or 20 million qualified Shorts views over the last 90 days, doubling the current 4,000 hours and 10 million views, while existing partners are grandfathered and Shorts revenue requires 10 million qualified views on a rolling 90 day window.

Dev News

YouTube Doubles What New Creators Need to Get Monetized

From February 1, 2027 the YouTube Partner Program asks new applicants for 8,000 watch hours or 20 million Shorts views. Existing partners are grandfathered.

Answer card describing Theseus Infrastructure, a platform announced on August 10 2026 by Anthropic, Macquarie Asset Management and GIC to develop, operate and lease data centre infrastructure to Anthropic as anchor tenant under long term agreements, with Macquarie funds and GIC owning the platform and funding the majority of equity per project, an initial focus on the United States, and a commitment from Anthropic to cover consumer electricity price increases.

Sysadmin News

Anthropic, Macquarie and GIC Form Theseus Infrastructure

A new platform will build, operate and lease data centres to Anthropic on long term agreements. Macquarie and GIC fund most of the equity per project.

Answer card describing FastFlowLM 1.0 as the first general release under the AMD ROCm organisation on GitHub, an MIT licensed runtime that executes language models on Ryzen AI XDNA2 neural processing units without a discrete GPU, in a 17 megabyte install supporting context lengths up to 256k tokens.

Dev News

FastFlowLM 1.0 Makes the Ryzen AI NPU a ROCm Target

FastFlowLM reached 1.0 and moved under the ROCm organisation on GitHub. It runs LLMs on Ryzen AI XDNA2 NPUs with no GPU, in a 17 MB install.

Answer card explaining that Fedora Workstation 44 ran slower than five other Linux distributions on a Framework Laptop 13 Pro with an Intel Core Ultra X9 388H, that updating Intel Thermald from 2.5.9 to 2.5.12 did not resolve it, and that replacing tuned-ppd with power-profiles-daemon restored competitive performance.

Sysadmin News

Fedora 44 Is Slow on Panther Lake, and Tuned Is Why

Phoronix traced Fedora Workstation 44's poor showing on an Intel Panther Lake laptop to tuned-ppd. Swapping in power-profiles-daemon closed most of the gap.

Answer card explaining that Ian Rogers of Google posted Linux kernel patches on August 11 2026 adding a mainline perf driver for the Broadcom AXI performance monitoring unit on Raspberry Pi 1 through 4, exposing uncore events under sysfs with CPU hotplug support and replacing out of tree DebugFS based drivers, with further work in progress for the BCM2712 in the Raspberry Pi 5.

Sysadmin News

Google Engineer Sends the Raspberry Pi AXI PMU to Mainline

Ian Rogers posted patches giving Raspberry Pi 1 through 4 a proper perf driver for the Broadcom AXI performance counters, replacing out of tree DebugFS hacks.

Answer card: the CachyOS August 2026 release published on August 9 adds experimental Server Edition installation profiles to the CLI installer, rewrites the chwd kernel manager back end and the Cachy Update tray applet from C plus plus to Rust, and moves the Shelly graphical package manager from C sharp to Zig, removing a managed runtime dependency.

Sysadmin News

CachyOS Preps a Server Edition and Rewrites Its Own Tools

The CachyOS August 2026 release adds experimental Server Edition install profiles, moves chwd and its updater applet to Rust, and rewrites Shelly in Zig.

Answer card: Linux 7.2-rc7 was released on August 9 2026 with roughly 400 non merge commits across 412 changed files, Linus Torvalds describing the size as the new normal driven by review from various AI tools, and the final Linux 7.2 expected around August 16 2026.

Sysadmin News

Linux 7.2-rc7 Is Big, and Torvalds Blames AI Review Tools

Linux 7.2-rc7 landed on August 9 with about 400 non merge commits across 412 files. Torvalds calls the volume the new normal, driven by AI review tools.

Answer card: Mesa 26.3 enables Large GRF mode on Intel DG2 Alchemist, Xe2 Battlemage and Lunar Lake integrated graphics, doubling the general register file slice per hardware thread from 128 to 256 registers of 64 bytes each, at the cost of dropping concurrent threads per execution unit from eight to four.

Dev News

Mesa 26.3 Doubles Intel GPU Registers, Halves the Threads

Mesa 26.3 turns on Large GRF mode for Intel Arc and Xe2 graphics. Shaders get 256 registers instead of 128, at the cost of half the hardware threads per EU.

Answer card: Intel announced on August 10 2026 an underwritten public offering of 15 billion dollars of common stock, with a 30 day underwriter option for a further 2.25 billion, proceeds going to general corporate purposes including capital expenditure and working capital, against a 2026 capital expenditure guidance of 20 billion dollars.

Sysadmin News

Intel Puts 15 Billion Dollars of New Stock on the Market

Intel announced a 15 billion dollar common stock offering on August 10, with a 2.25 billion option on top, citing unprecedented investment in AI compute.

Answer card: the Linux Mint July 2026 monthly news moves kernel management from Update Manager into the System Administration tool for both Linux Mint and LMDE, replaces individual kernel selection with kernel series tracking plus a retention count, removes outdated kernels automatically once a week with a history view, adds a page for setting system wide environment variables, and ships new hardware enablement ISOs carrying kernel 7.0 for Linux Mint 22.3.

Sysadmin News

Linux Mint Moves Kernel Cleanup Into System Administration

Linux Mint moved kernel management into System Administration, switched to tracking kernel series, and added a system wide environment variables page.

Answer card: Meta Superintelligence Labs published Muse Glimmer on August 10 2026, a 30 billion parameter dense model under an Apache 2.0 licence, distilled from Muse Spark, tuned for local agent workflows with tool calling and failure recovery, running in about 17 GB of memory at 4 bit quantisation with a context window of 131072 tokens by default and 262144 at most.

Dev News

Meta Ships Muse Glimmer, a 30B Agent Model for One GPU

Meta published Muse Glimmer on August 10: 30 billion dense parameters, Apache 2.0, up to 262k context, and about 17 GB of VRAM at 4 bit.

Answer card titled write it once, ship it anywhere, explaining that an Agent Plugins package is a directory containing a plugin.json manifest, a skills folder and an mcp.json file, that the core maintainers are Amazon, Cursor, Microsoft, OpenAI, Vercel and Google since August 6, and highlighting 1.0.0 as the specification version.

Dev News

Agent Plugins 1.0 Makes Skills and MCP Servers Portable

Amazon, Cursor, Microsoft, OpenAI, Vercel and now Google back Agent Plugins 1.0, a directory format that packages Agent Skills and MCP servers once for any client.

Answer card titled humans as a rounding error, explaining that Cloudflare chief financial officer Thomas Seifert told analysts on the second quarter 2026 earnings call that non-human traffic could reach one thousand times human traffic within five years, that he added he has called it wrong at every point so far, and highlighting May 2026 as the month machines overtook humans.

Network News

Cloudflare Sees Machine Traffic Hitting 1000x Human Levels

Cloudflare's CFO told investors non-human traffic could reach 1,000 times human traffic within five years, after machines overtook humans on its network in May 2026.

Answer card titled cluster spreading that skips small cores, explaining that on Intel hybrid parts the misfit tasks reach the performance cores but the leftovers fail to spread across the efficiency core clusters, that the fix is queued in tip sched core for Linux 7.3 in August 2026, and highlighting 2021 as the year cluster scheduling was added to the kernel.

Sysadmin News

Linux 7.3 Fixes Cluster Load Balancing on Intel Hybrid CPUs

Intel patches queued in the tip tree for Linux 7.3 repair cluster scheduling on hybrid CPUs, so leftover tasks spread across E core clusters instead of bunching up.

Answer card titled bought months before anyone said so, explaining that NextSlide turned notes and documents into polished editable presentations, that the acquisition closed earlier in 2026 and surfaced publicly on August 8, that the team now works on ChatGPT, and highlighting 2026 as the year the deal closed.

Dev News

OpenAI Quietly Bought NextSlide, an AI Presentation Startup

OpenAI acquired NextSlide earlier in 2026 and only confirmed it on August 8. Terms were undisclosed and the team now works on ChatGPT. Here is what it signals.

Answer card titled one extra day of cyclone warning, explaining that Google DeepMind open sourced WeatherNext Cyclones and WeatherNext 2 on August 6, 2026, with code under Apache 2.0 and weights under Creative Commons Attribution 4.0, and highlighting 100 km as the reduction in three day track position error against the ECMWF ensemble.

Dev News

DeepMind Open Sources WeatherNext, Its Cyclone Model

Google DeepMind released the WeatherNext code under Apache 2.0 and the weights under CC BY 4.0, with a mini variant that runs in a free Colab notebook.

Answer card titled a sensor that was already supported, explaining that the Dell Latitude 7320 Detachable front camera never worked on Linux because its OV5678 sensor had no driver, that it turned out to report the chip id of the already supported OV5675, and highlighting five years as the time the laptop went without a working camera.

Sysadmin News

Dell Latitude 7320 Gets Its Linux Webcam After Five Years

Patches on the linux-media list finally enable the front camera on the Dell Latitude 7320 Detachable, after someone worked out the OV5678 is a relabelled OV5675.

Answer card titled open weight models for science, explaining that the US Department of Energy launched the Genesis Open Models Initiative on August 7, 2026, that the first model is Genesis-Science-1 built with Arcee, and highlighting August 14 as the closing date for pretraining data contributions.

Dev News

DOE Launches Genesis Open Models, With Deadlines in August

The US Department of Energy is building open weight foundation models for science, starting with Genesis-Science-1, and it is taking contributions until late August.

Answer card: AMD announced on August 6 2026 that it is acquiring Taalas, a Toronto startup founded in 2023 whose model specific integrated circuits etch model weights directly into silicon instead of storing them in high bandwidth memory, with its HC1 test chip on TSMC 6 nanometre serving Llama 3.1 8B at 16,960 tokens a second.

Sysadmin News

AMD Buys Taalas, Which Etches Model Weights Into Silicon

AMD is acquiring Toronto startup Taalas, whose chips burn model weights into the die instead of HBM. Its HC1 part served Llama 3.1 8B at 16,960 tokens a second.

Answer card: Linux 7.3 is picking up device tree support for Apple M3 Pro, M3 Max and M3 Ultra from a pull request by Sven Peter, bringing them to the same baseline the plain M3 reached in Linux 7.2, with no GPU acceleration yet, plus HWMON sensor support for M1 and M2 series hardware.

Sysadmin News

Linux 7.3 Picks Up M3 Pro, Max and Ultra Device Trees

Sven Peter has queued device tree support for Apple's M3 Pro, Max and Ultra in Linux 7.3, plus HWMON sensors for M1 and M2. It boots. It does not yet have a GPU.

Answer card: NVIDIA released NOOA on August 7 2026 under Apache 2.0, a model agnostic Python framework where an AI agent is a single Python class, methods are actions, docstrings are prompts, type annotations are contracts, and a method body of three dots is filled in at runtime by the model.

Dev News

NVIDIA's NOOA Makes an AI Agent a Single Python Class

NVIDIA open sourced NOOA under Apache 2.0, a framework where an agent is one Python class: methods are actions, docstrings are prompts, types are contracts.

Answer card: Linux 7.3 changes TTM so that a protected device memory cgroup allocation evicts unprotected buffers from VRAM instead of falling back to system memory, work by Natalie Vock of Valve's Linux graphics team after eight rounds of revision, applying to AMDGPU, Xe and Nouveau and also removing an artificial 64 megabyte limit in UDMABUF.

Dev News

Linux 7.3 Lets Apps Claim the VRAM They Were Promised

Valve's Natalie Vock got TTM to evict unprotected buffers instead of falling back to system memory, so cgroup VRAM guarantees finally mean something.

Answer card explaining that the Financial Times reported on August 7 2026 that ByteDance is in early pre-training on a model of up to 10 trillion parameters, more than three times the roughly 2.8 trillion of Moonshot Kimi K3, with industry estimates placing Anthropic Mythos 5 near 8 trillion and Fable 5 near 5 trillion, and noting that pre-training typically runs three to six months with no confirmed release date.

Dev News

ByteDance Is Training a 10 Trillion Parameter Model

The Financial Times reports ByteDance is in early pre-training on a model of up to 10 trillion parameters. Here is what that number does and does not tell you.

Answer card explaining that Lumilens emerged from stealth on August 6 2026 with more than 900 million dollars in total funding including a Series C above 700 million dollars at a 5.51 billion dollar valuation, founded in early 2024 by Ankur Singla, building the LumiCore platform of silicon photonics, mixed signal integrated circuits, electrical optical interposers and optical systems, shipping near package and co-packaged optics for scale up networks and 800G and 1.6T pluggable transceivers for scale out.

Network News

Lumilens Leaves Stealth With 900 Million Dollars for Optics

Lumilens came out of stealth on August 6 with a 700 million dollar Series C, a 5.51 billion valuation, and co-packaged optics already shipping to a hyperscaler.

Answer card explaining that Triton is a DirectX 11 driver for QEMU Windows guests that implements the Windows Device Driver Interface as a user mode display driver plus a kernel mode driver, carrying Direct3D commands over VirtIO through the Neptune protocol to virglrenderer on the host, with Claude Opus 5 and Claude Fable 5 assisting development.

Sysadmin News

Triton Brings DirectX 11 to QEMU as a Real Windows Driver

Triton implements the DirectX Device Driver Interface inside a QEMU Windows guest instead of swapping DLLs. Claude Opus 5 and Fable 5 helped write it.

Answer card summarising Wine 11.15, released Saturday August 8 2026 with 41 bug fixes, ARM64EC build support through MinGW, additional key derivation function algorithms in BCrypt, more format conversions in WindowsCodecs, two Wayland corrections for double sRGB conversion and offset 4 by 3 fullscreen content, and the closure of bug 4811 against MSXML3 which was filed in March 2006.

Dev News

Wine 11.15 Closes an MSXML3 Bug Filed in March 2006

Wine 11.15 lands 41 fixes, adds ARM64EC builds through MinGW, corrects washed out colours on Wayland, and resolves bug 4811 after twenty years.

Answer card: Btrfs removed its copy on write fixup worker during the Linux 7.2 merge window on the assumption that the memory management layer had solved the underlying conditions, which instead produced silent data loss, so David Sterba of SUSE restored over 600 lines of the infrastructure in a pull request accepted for Linux 7.2-rc7.

Sysadmin News

Btrfs Restores Its COW Fixup Worker After Silent Data Loss

Btrfs dropped its COW fixup worker in the Linux 7.2 merge window, assuming memory management had solved the problem. It had not, and 600 lines are back in rc7.

Answer card: Bloomberg reported on August 6, 2026 that OpenAI's first hardware device is a doughnut shaped smart speaker about the size of a hockey puck, with no display, microphones and speaker grilles, a camera and environmental sensors, a battery for portability, moving parts and lights that signal listening and responding, a premium metal body designed by Jony Ive, and a price between three hundred and four hundred dollars, with a full launch in 2027.

Dev News

OpenAI's First Device Reported as a $300 Doughnut Speaker

Bloomberg reports OpenAI's first hardware product is a hockey puck sized doughnut shaped speaker with no display, a camera, moving parts and a price above $300.

Answer card: a request for comments from Juan Yescas at Google proposes Page Alloc Hogger, a DebugFS interface that allocates memory pages from a chosen NUMA node, zone, migration type and allocation order, so developers can reproduce low memory conditions and inspect direct reclaim, kswapd, the OOM killer and allocation fallbacks without writing a custom kernel driver.

Sysadmin News

Page Alloc Hogger Aims Linux Memory Pressure via DebugFS

A Google RFC adds a DebugFS interface that allocates pages from a chosen node, zone, migration type and order, so you can reproduce low memory conditions on demand.

Answer card: AMD open sourced Spur on August 7 2026, a GPU first job scheduler written in Rust under Apache 2.0, drop in compatible with the Slurm command line, REST API and C FFI, using embedded Raft consensus for high availability instead of an external database.

Sysadmin News

AMD Ships Spur, a Rust Job Scheduler That Speaks Slurm

AMD open sourced Spur, a GPU first job scheduler written in Rust that is drop in compatible with Slurm commands, eight months after NVIDIA bought SchedMD.

Answer card: GCC 16.2 was released on August 7 2026 with 102 known bug fixes over GCC 16.1, described upstream as regression fixes and documentation only, with no new language features or optimisation changes, making it the recommended build of the GCC 16 series.

Dev News

GCC 16.2 Lands With 102 Fixes and Nothing Else

GCC 16.2 shipped on August 7 with 102 bug fixes over 16.1 and no new features. That is exactly what a point release should be, and it is the one to build with.

Answer card: PoCL 7.2-RC1 released on August 7 2026 achieved official OpenCL 3.0 conformance for its CPU device on both RISC-V and x86-64, validated with the Khronos OpenCL conformance test suite tag v2026-03-25-00, tested on the Star64 and Milk-V Jupiter RISC-V boards and an AMD Ryzen 9 9900X.

Dev News

PoCL 7.2 Makes RISC-V a Conformant OpenCL 3.0 Target

PoCL 7.2-RC1 passed the Khronos conformance suite on RISC-V hardware as well as x86_64, so a Star64 or Milk-V Jupiter now runs certified OpenCL 3.0 on its CPU.

Answer card: Mesa 26.2 was released on August 5 2026 with VK_EXT_mesh_shader support in the NVK driver for NVIDIA hardware, more than sixty new Vulkan and OpenGL extensions across the tree, OpenCL 3.1 in five Rusticl drivers, Vulkan 1.4 exposed by the KosmicKrisp Metal translation layer and a new KRAID compiler for Arm Mali.

Dev News

Mesa 26.2 Brings Mesh Shaders to the NVK Driver

Mesa 26.2 shipped on August 5 with VK_EXT_mesh_shader in NVK, OpenCL 3.1 across five Rusticl drivers, and Vulkan 1.4 exposed by the Metal translation layer.

Answer card: Meta launched Muse Code in beta on August 5 2026, a terminal coding agent for macOS and Linux built on Muse Spark 1.2, which spawns sub agents in isolated git worktrees and offers a contributor pricing tier roughly ten times cheaper in exchange for permission to train on your prompts and completions.

Dev News

Meta Ships Muse Code, a Terminal Agent for Large Repos

Meta launched Muse Code on August 5, a terminal coding agent for macOS and Linux, with a contributor tier that trades your prompts for a far lower price.

Answer card: Rust Coreutils 0.10 was released on August 5 2026 with 645 GNU test suite cases passing against 29 failures, a pass rate of 93.48 percent up from 90.58 percent in version 0.9, an anti panic campaign across more than twenty utilities, new options including mv --exchange and rm --one-file-system, and a new wasm32-wasip2 build target.

Sysadmin News

Rust Coreutils 0.10 Clears 93 Percent of the GNU Tests

Rust Coreutils 0.10 shipped on August 5 with 645 of the GNU test suite passing, an anti panic campaign across 20 utilities, and mv --exchange for atomic swaps.

Answer card: a patch queued in the thunderbolt tree for Linux 7.3 adds a busy poll ConfigFS attribute to USB4STREAM, switching data transfer rings from interrupt driven delivery to polling for lower host to host latency, at the cost of continuous CPU consumption and the loss of poll(2) support on the interface.

Network News

Linux 7.3 Adds a Busy Poll Switch to USB4 Data Streams

A ConfigFS knob queued for Linux 7.3 lets USB4STREAM transfer rings run in polling mode instead of interrupts, trading CPU cycles for lower host to host latency.

Answer card: HP, Asus and Acer began fitting DRAM from China's CXMT in notebooks in limited volumes during 2026, after validation completed mid year, restricted to models sold outside the United States, against a backdrop of DRAM prices tripling in eighteen months and AI data centres absorbing about seventy percent of memory production.

Sysadmin News

HP, Asus and Acer Start Shipping Chinese DRAM in Laptops

Three of the largest PC makers have begun fitting CXMT memory in notebooks sold outside the US. Here is what that means for the machines you buy this year.

Answer card: on August 5 2026 Jeff Dean left Google after 27 years alongside Sanjay Ghemawat, Oriol Vinyals and Quoc Le to co-found Discovery Loop, a public benefit corporation backed by Alphabet, Radical Ventures and Khosla Ventures, while Demis Hassabis became Alphabet chief scientist and Koray Kavukcuoglu took over Google DeepMind.

Dev News

Jeff Dean Quits Google for Discovery Loop After 27 Years

Jeff Dean, Sanjay Ghemawat, Oriol Vinyals and Quoc Le are leaving Google for Discovery Loop, a startup Alphabet is funding. Why the infrastructure world should care.

Answer card: Qualcomm posted an RFC on August 6 2026 proposing Synx, a SoC wide synchronization framework that coordinates fences between Linux clients and non Linux execution environments such as firmware and remote processors, designed to complement rather than replace dma-fence and sync_file.

Sysadmin News

Qualcomm Proposes Synx, a Cross SoC Fence Layer for Linux

Qualcomm posted an RFC on August 6 to upstream Synx, a synchronization framework that lets Linux share fences with firmware and remote processors on a SoC.

Answer card explaining that on August 4 2026 AMD's Krishnamoorthi M posted Linux kernel patches proposing eSPI as its own new bus type rather than an extension of the SPI subsystem, with an AMD eSPI driver as the first user, tested on an AMDI0070 platform, and with ACPI only enumeration as a current limitation.

Sysadmin News

AMD Wants eSPI to Be Its Own Bus Type in the Linux Kernel

AMD posted patches on August 4 proposing eSPI as a new Linux bus type rather than an extension of the SPI subsystem, with an AMD driver as its first user.

Answer card explaining that C-Kermit 11.0.506 was released on August 3 2026 as the first non beta release since 9.0.302 in 2011, dedicated to Frank da Cruz, adding IPv6 support, more than 1700 test cases and continuous integration on Linux, macOS, FreeBSD and NetBSD, while shrinking the codebase by roughly 2600 lines and changing several long standing defaults.

Network News

C-Kermit 11 Ships, Fifteen Years After the Last One

C-Kermit 11.0.506 landed on August 3 with IPv6, 1700 test cases and CI on four systems. Several defaults changed, so old scripts need a read before you upgrade.

Answer card explaining that the Linux Foundation launched the Tokenomics Foundation on August 4 2026 with about thirty founding members including IBM, Oracle, SAP, ServiceNow, Broadcom, Lenovo, Accenture, JPMorganChase and BNY, to define vendor neutral standards for measuring AI cost and return on investment, working in partnership with the FinOps Foundation and adding token cost telemetry to the FOCUS billing specification.

Sysadmin News

The Linux Foundation Takes On the AI Bill Nobody Can Read

The Tokenomics Foundation launched on August 4 with about 30 members including IBM, Oracle, SAP and JPMorganChase, aiming to standardise how AI cost is measured.

Answer card explaining that a Linux process builder patch series posted by Li Chen and covered on August 4 2026 proposes creating an empty process with pidfd_open, configuring it incrementally with a new pidfd_config system call, then executing, so that posix_spawn no longer has to be built on fork or vfork.

Dev News

Linux May Get a Process Builder Instead of fork()

Li Chen posted a process builder patch series covered on August 4, following Christian Brauner's idea of building a process through pidfd instead of fork().

Answer card: Apple has moved for a preliminary injunction against OpenAI and two named former Apple employees in its trade secrets case over consumer hardware, seeking expedited discovery and depositions from OpenAI and IO Products, with a hearing scheduled for October 1 2026 and OpenAI publicly disputing the allegations.

Dev News

Apple Asks a Court to Freeze OpenAI's Hardware Work For Now

Apple filed for a preliminary injunction against OpenAI and two former employees in its trade secrets case. The hearing is set for October 1, 2026.

Answer card: on August 5 2026 Cloudflare released Cloudflare OS under the Apache 2.0 licence, an agent workspace built on Cloudflare Workers that runs server code in Dynamic Workers with outbound networking disabled and gives every app instance its own Durable Object Facet backed by a private SQLite database.

Dev News

Cloudflare Open Sources Its Internal Agent Workspace

Cloudflare OS went public under Apache 2.0 on August 5. It runs on Workers, Durable Object Facets and Cap'n Web, and the README says workerd alone is enough.

Answer card: FEX 2608, released August 4 2026, implements spinloops under Wine using the ARM WFE instruction so the core enters a low power state while waiting on an atomic, adds 256-bit SVE improvements, fixes AVX signal state save and restore, and removes the deprecated FEXInterpreter binary after nearly a year.

Dev News

FEX 2608 Makes ARM Chips Idle Properly While Emulating x86

FEX 2608 landed on August 4 with WFE based spinloops under Wine, 256-bit SVE work, AVX signal state fixes, and the removal of the old FEXInterpreter binary.

Answer card: a patch by Rafael Wysocki makes the Linux intel_pstate driver write zero into the desired performance field of MSR_HWP_REQUEST on Intel processors with Dynamic Efficiency Control, so the processor firmware chooses the operating point itself, targeting the Linux 7.3 merge window with no published benchmark figures.

Sysadmin News

Linux Stops Telling Panther Lake Chips What Speed to Run

A patch from Rafael Wysocki makes intel_pstate leave the desired performance field at zero on chips with Dynamic Efficiency Control, targeting Linux 7.3.

Answer card: FFmpeg 9.0 was released on August 3 2026 with an animated WebP decoder and demuxer, an APV Vulkan hardware acceleration path, four AMD AMF additions and an ONNX Runtime DNN backend, while removing CELT decoding, ogg CELT parsing and deprecated NVENC options.

Dev News

FFmpeg 9.0 Adds Animated WebP and Drops CELT Decoding

FFmpeg 9.0 landed on August 3 with animated WebP, an APV Vulkan hwaccel, four new AMD AMF filters and an ONNX Runtime DNN backend. Three removals can break builds.

Answer card: Felipe Borges published a beta of the rewritten GNOME Boxes on August 3 2026 after two years of work, moving the application to GTK4 and libadwaita, replacing the GTK3 SPICE widget with libmks, shipping Flatpak only, installing Windows 11 without manual Secure Boot or TPM workarounds, and adding VSOCK and port forwarding.

Sysadmin News

GNOME Boxes Beta: A Two Year Rewrite, and Flatpak Only

Felipe Borges published a beta of the rebuilt GNOME Boxes on August 3. It installs Windows 11 without workarounds, adds VSOCK and port forwarding, and is Flatpak only.

Answer card: Alibaba released Qwen3.8-Max on August 3 2026 with 2.4 trillion total parameters, a one million token context window, pricing of two dollars per million input tokens and six dollars per million output tokens, and open weights promised for the following week alongside a Qwen3.8-27B model.

Dev News

Qwen3.8-Max: 2.4 Trillion Parameters, Open Weights Promised

Alibaba's biggest model yet ships on the API at $2 per million input tokens, with open weights and a 27B companion promised for the following week.

Answer card: on August 4 2026 Anthropic signed a six year 10 billion dollar compute agreement with Volta Infra Holdings for 133 megawatts of NVIDIA Vera Rubin capacity at a hydroelectric powered Norwegian data center operated by Bitdeer, with phased delivery in late 2026 and early 2027.

Sysadmin News

Anthropic Signs a 10 Billion Dollar Compute Deal With Volta

Anthropic committed 10 billion dollars over six years to Volta Infra for 133 MW of Vera Rubin capacity at a hydro powered Norwegian site operated by Bitdeer.

Answer card: on August 4 2026 Greg Kroah-Hartman announced that the Linux kernel drivers/staging subsystem will reject LLM generated patches, because staging exists to teach new kernel developers, with a single exception for genuine fixes that the submitter has tested on the real hardware and described how.

Dev News

Linux Staging Tree Now Rejects LLM Generated Patches

Greg Kroah-Hartman closed drivers/staging to LLM written patches on August 4, keeping one narrow exception, because staging exists to teach new kernel developers.

Answer card: on August 4 2026 NVIDIA became the fourth OEM premier sponsor of the Linux Vendor Firmware Service, after Dell, Lenovo and HP, with DGX Spark firmware already being distributed through fwupd and the project reaching the funding target it set the previous year.

Sysadmin News

NVIDIA Joins Dell, Lenovo and HP in Funding the LVFS

NVIDIA joined Dell, Lenovo and HP as an LVFS premier sponsor on August 4, DGX Spark firmware now ships through fwupd, and the project hit its funding target.

Answer card: Wild linker 0.10 was released on August 4 2026 with parallelised --gdb-index support that speeds up GDB startup, a new default that avoids memory mapping the output file on btrfs and vfat filesystems, extensive linker script additions, initial 32 bit work and in memory file support in libwild.

Dev News

Wild Linker 0.10 Adds gdb-index and Skips mmap on Btrfs

Wild 0.10 shipped on August 4 with a parallelised --gdb-index implementation, faster debugger startup, and a new default that avoids mmap on btrfs and vfat.

Answer card: AerynOS 2026.08 deploys Versioned Repositories phase 2 so the moss package manager can cross breaking on disk format changes during a normal update, and adds experimental OpenZFS for data drives, systemd 261.2 and Linux 7.1.5.

Sysadmin News

AerynOS Can Now Update Across a Breaking Format Change

AerynOS 2026.08 lands Versioned Repositories phase 2, so moss crosses breaking on disk format changes during a normal update. Plus reflink on XFS and OpenZFS.

Answer card: Kakehashi is a Rust userspace translation layer that loads Mach-O binaries on Linux aarch64, maps a freestanding libSystem and translates BSD system calls, running 7-Zip and curl at around 5.2 times slower than native on runners costing roughly a twelfth of macOS.

Dev News

Kakehashi Runs macOS ARM64 Binaries on Linux, No VM

Kakehashi is a Rust userspace layer that loads Mach-O binaries on Linux aarch64 and translates BSD syscalls, aimed at running Darwin CLI tools on cheap CI runners.

Answer card: Microsoft named memory optimization for 8 GB PCs and above a Windows 11 priority for the rest of 2026, covering a more efficient memory allocator, WinUI 3 tuning and lighter Chromium and WebView2 components, with broad rollout starting this fall.

Sysadmin News

Microsoft Puts Windows 11 on a Memory Diet for 8 GB PCs

Microsoft named memory optimization for 8 GB PCs a Windows 11 priority: a leaner allocator, WinUI 3 tuning and lighter WebView2, rolling out broadly this fall.

Answer card: Mutter 51 beta, tagged on August 2 2026, adds ext-background-effect-v1 background blur, switches to the upstream xdg_session_management_v1 protocol, and lands frame scheduling improvements, ahead of the GNOME 51 A Coruna release on September 16.

Sysadmin News

GNOME 51 Beta Brings Real Background Blur to Wayland

Mutter 51 beta adds ext-background-effect-v1, so apps can finally ask the compositor to blur what is behind them. Plus frame scheduling work and session restore.

Answer card: Linux 7.3 adds mainline support for the Realtek RTL8261C and RTL8261D 10 Gigabit PHYs, parts that draw 1.65 watts in an eight by eight millimetre package and target WiFi 6 and 7 routers, switches, PON and cable modems, replacing an out of tree driver that was archived in June 2026.

Network News

Linux 7.3 Picks Up Realtek's Cheap 10 Gigabit PHYs

Mainline support for the RTL8261C and RTL8261D is queued for Linux 7.3. These are the 1.65 W parts meant to put 10GbE in routers and switches you can afford.

Answer card: the MacBook Air now ships in two to six weeks after a two hundred dollar price rise in June, because DRAM makers are diverting capacity to high bandwidth memory for AI accelerators, with TrendForce forecasting server DRAM contract prices up thirteen to eighteen percent in the third quarter of 2026.

Sysadmin News

The Memory Shortage Has Reached Apple's Best Selling Mac

Bloomberg reports MacBook Air ship dates slipping to six weeks as AI data center demand drains DRAM supply. What it means for anyone buying memory this year.

Answer card: Article 50 of the EU AI Act applies from August 2, 2026, covering chatbot disclosure, machine readable marking of synthetic content, emotion recognition notice and deepfake labelling, with fines up to 15 million euros or 3% of worldwide turnover.

Dev News

EU AI Act: Chatbot and Deepfake Disclosure Rules Are Live

Article 50 of the EU AI Act applies from August 2, 2026: chatbot disclosure, machine readable marking of AI output, deepfake labels, fines up to 15 million euros.

Answer card: an umbrella Go proposal filed July 28, 2026 by Alan Donovan targets Go 1.28 with container/set, container/hash, container/ordered, container/mapset and container/heap/v2, built on the maphash.Hasher interface that shipped in Go 1.27.

Dev News

Go Wants Sets and Ordered Maps in the Standard Library

A Go working group filed an umbrella proposal for container/set, container/hash, container/ordered and heap/v2, targeting Go 1.28 and built on maphash.Hasher.

Answer card: NetBSD 11.0 is the nineteenth major release, adding a 64-bit RISC-V port for VisionFive 2, STAR64 and QEMU, a MICROVM x86 kernel that boots in about 10 milliseconds, Qualcomm Snapdragon X Elite support and eleven new Linux system calls.

Sysadmin News

NetBSD 11.0 Adds RISC-V and a Kernel That Boots in 10 ms

NetBSD 11.0 ships a 64-bit RISC-V port, a MICROVM kernel booting in about 10 ms, Snapdragon X Elite support, eleven new Linux syscalls and layer 2 npf filtering.

Answer card: Box64 0.4.4, released August 2 2026, enables DynaCache by default with a 2 GB compressed cap, defaults BOX64_DYNAREC_CALLRET to 2, and adds a Python based box64-configurator for per application profiles on ARM64, RISC-V and LoongArch.

Dev News

Box64 0.4.4 Turns Its Code Cache On by Default

The x86_64 emulator for ARM64, RISC-V and LoongArch now enables DynaCache out of the box, capped at 2 GB, and ships a graphical per application configurator.

Answer card: GNOME is discussing a formal RFC process proposed by Sophie Herold on July 27, 2026, with a fourteen day final concern period, alongside a reference document teaching LLMs how to write GNOME Shell extensions correctly.

Dev News

GNOME Drafts an RFC Process as AI Extensions Flood Review

Sophie Herold proposed a formal RFC process for GNOME decisions, while a separate reference document tries to teach LLMs how to write a Shell extension properly.

Answer card: Intel Starfire is a space grade Panther Lake derived SoC on Intel 18A with a three tile NPU rated up to 75 TOPS, and its first Linux patch adds a device ID to the EDAC IGEN6 driver for kernel 7.3.

Sysadmin News

Intel's Starfire Space SoC Gets Its First Linux Patch

Starfire pairs Intel 18A cores with a three tile NPU rated up to 75 TOPS, and its first upstream Linux patch just landed in the EDAC subsystem for 7.3.

Answer card: a fix merged for Linux 7.2-rc6 reassigns AMD Family 1Ah models 0xd0 to 0xd7 back to Zen 5, leaving Zen 6 with 0x50 to 0x5f, 0x80 to 0xaf, 0xc0 to 0xcf and 0xd8 to 0xef.

Sysadmin News

Linux 7.2 rc6 Stops Filing Some Zen 5 Chips Under Zen 6

A May patch handed models 0xd0 to 0xef to Zen 6. An x86 fix merged for Linux 7.2-rc6 gives 0xd0 to 0xd7 back to Zen 5 before the kernel ships.

Answer card: Dasharo v0.9.0, released July thirtieth 2026 by 3mdeb, is the first open source firmware for a modern AMD Ryzen AM5 desktop board, the MSI PRO B850-P WIFI, built on coreboot 25.12 and AMD openSIL with 79.1% less closed source code than the stock MSI BIOS.

Sysadmin News

Open Source Firmware Ships for a Modern AMD AM5 Board

3mdeb released Dasharo v0.9.0 for the MSI PRO B850-P WIFI, built on coreboot 25.12 and AMD openSIL, with 79.1% less closed source code than the stock BIOS.

Answer card: Google launched a Nano Banana 2 create image feature in Google Earth on July thirtieth 2026 and rolled it back within 48 hours after researchers showed fabricated scenes could be pinned to real coordinates, with the SynthID watermark defence shown to be bypassable.

Dev News

Google Pulls Its Earth AI Image Tool One Day After Launch

Google rolled back the Nano Banana 2 create image feature in Google Earth within 48 hours, and the SynthID watermark defence is the part worth studying.

Answer card: Servo 0.4.0 was released on July thirty first 2026 with a record 558 commits and 20 new contributors, adding SharedWorker, nine media queries, pointer capture and touch events, with lichess.org, Google Photos and Zulip now rendering correctly.

Dev News

Servo 0.4 Ships a Record 558 Commits and Real Site Compat

Servo 0.4.0 landed on July thirty first with 558 commits, 20 new contributors, SharedWorker, nine new media queries, and lichess and Google Photos rendering.

Answer card: on July thirty first 2026 the 42nd civil chamber of Munich Regional Court I ruled that six GEMA represented songs were reproducibly memorised inside Suno models 3.5 and 4, infringing both the reproduction right and the right of making works available to the public, with the text and data mining exception rejected.

Dev News

German Court Rules Songs Inside Model Weights Are Copies

Munich Regional Court I ruled that Suno infringed copyright by memorising six GEMA songs inside its 3.5 and 4 models, and rejected the text and data mining defence.

Answer card: KDE Plasma 6.8 adds linux-dmabuf version 6 support to KWin, letting the compositor advertise multiple GPUs so buffers no longer round trip through system memory. On a Framework Laptop 13 with an external RX 5700 XT, vkcube rose from 55 to 120 frames per second and Cyberpunk 2077 on low settings from 27 to 50 frames per second.

Sysadmin News

KWin Stops Copying Frames Between GPUs in Plasma 6.8

Xaver Hugl's linux-dmabuf v6 work lets KWin advertise every GPU instead of one. vkcube went from 55 to 120 fps and Cyberpunk 2077 from 27 to 50 fps on an eGPU.

Answer card: Linux 7.2-rc6 adds libata NOLPM quirks that disable SATA Link Power Management for three Western Digital drives, the 10 TB WDC WD100EFGX-68CPLN0 and WDC WD102KFBX-68M95N0 NAS disks and a WD Green 2.5 inch 480 GB SSD, on which the power saving feature takes the SATA link offline. The workaround on older kernels is the libata.force=nolpm boot parameter.

Sysadmin News

Linux 7.2 Turns Off SATA Power Saving on Three WD Disks

Two libata quirks merged for Linux 7.2-rc6 turn off Link Power Management on two 10 TB WD Red class NAS drives and a WD Green 480 GB SSD that drop off the bus.

Answer card: the European Commission opened the AI Gigafactories call on July 30, 2026, with up to seven sites, three large facilities of at least 100,000 chips and four smaller ones of at least 75,000, ten billion euros of public funding, a November 12 bid deadline and construction starting in early 2027.

Sysadmin News

EU Opens Bidding for Seven AI Gigafactories

The European Commission opened its AI Gigafactories call on July 30, 2026: up to seven sites, 75,000 to 100,000 chips each, bids close November 12.

Answer card: Google DeepMind released Gemini Robotics 2 on July 30, 2026 as three models, an ER 2 reasoning model available in the Gemini API, a whole body vision language action model for early access partners, and an on-device model that adapts to a new robot with fewer than 200 examples.

Dev News

Gemini Robotics 2: Three Models, One Runs on the Robot

Google DeepMind shipped Gemini Robotics 2 on July 30, 2026: a whole body action model, an ER 2 reasoning model in preview, and an on-device variant.

Answer card: the Linux Foundation Open Programmable Infrastructure project released OPI Abstraction v0.1.0 on July 29, 2026, a vendor neutral API layer across 26 repositories covering Kubernetes integration, provisioning and observability, plus a Kubernetes Network Function Offload Blueprint built with F5 NGINX, Intel and Red Hat.

Network News

OPI Ships Its First Release to Standardise DPU and IPU APIs

The Linux Foundation's Open Programmable Infrastructure project published OPI Abstraction v0.1.0, a vendor neutral API layer across 26 repositories.

Answer card: a Nouveau patch by David Airlie staged for Linux 7.3 allows allocating NVDEC channels via ABI16, letting Mesa's NVK driver advertise Vulkan Video H.264 decode on Turing and newer NVIDIA GPUs.

Dev News

Nouveau Gains NVDEC Channels, Unlocking Vulkan Video in NVK

A Nouveau patch queued for Linux 7.3 lets userspace allocate NVDEC channels through ABI16, which is what Mesa's NVK driver needs to advertise H.264 decode.

Answer card: Nscale agreed on July 30, 2026 to acquire Anyscale, moving about 200 staff and the managed Ray platform into an operator with more than one gigawatt of capacity, while Ray itself stays governed by the PyTorch Foundation.

Sysadmin News

Nscale Buys Anyscale, the Company Behind Ray

Nscale agreed on July 30, 2026 to acquire Anyscale, adding an orchestration layer above its GPU fleet. Ray itself stays governed by the PyTorch Foundation.

Answer card: OpenAI cut GPT-5.6 Luna to 20 cents per million input tokens and $1.20 per million output tokens on July 30, 2026, an 80% reduction, while Terra fell 20% to $2 and $12 and the flagship Sol stayed at $5 and $30.

Dev News

OpenAI Cuts GPT-5.6 Luna 80% Three Weeks After Launch

OpenAI cut GPT-5.6 Luna API prices by 80% and Terra by 20% on July 30, 2026, three weeks after the family reached general availability. Sol is unchanged.

Answer card: Kuiper Systems asked the FCC to authorise up to 5,105 Amazon Leo satellites in five shells between 510 and 580 km, using L-band and S-band for device links and Ka-band and V-band for feeder links, with deployment beginning in 2028.

Network News

Amazon Leo Files for 5,105 Direct to Phone Satellites

Amazon asked the FCC to authorise up to 5,105 satellites for direct to device service on Globalstar L-band and S-band spectrum, with deployment starting in 2028.

Answer card: GlobalFoundries signed a $300 million CHIPS research letter of intent for silicon photonics on July 29, 2026, targeting 400Gb/s and a 5x energy efficiency gain with its SCALE co-packaged optics platform, with the US Department of Commerce taking about 1% equity.

Network News

GlobalFoundries Gets $300M to Move Optics Into the Package

A $300 million CHIPS award funds GlobalFoundries work on co-packaged optics, targeting 400Gb/s and a 5x energy efficiency gain, with Washington taking 1% equity.

Answer card: Azure surpassed $100 billion in annual revenue in fiscal 2026 and grew 43% in the fourth quarter, while Microsoft spent about $41 billion on capital expenditure and finance leases in the quarter and guided above $50 billion for the next one.

Sysadmin News

Azure Passes $100B a Year and Still Runs Short of Capacity

Azure crossed $100 billion in annual revenue and grew 43% last quarter, yet Microsoft's CFO says cloud demand still exceeds the capacity available to serve it.

Answer card: YSERVER 1.4.0 released July 29, 2026, a from scratch X11 server in Rust that now gives Chrome and Steam GPU acceleration, runs full screen games under Cinnamon, and passes 672 XTS5 tests across AMD, Intel, NVIDIA and Apple Silicon.

Dev News

YSERVER 1.4 Is a Rust X11 Server That Now Runs Steam

YSERVER 1.4.0 gives Chrome and Steam real GPU acceleration on a from scratch Rust X11 server, written largely with Claude Code and passing 672 XTS5 tests.

Answer card: AMD posted Linux kernel patches on July 29, 2026 adding Zen 6 client support to the amd_hsmp driver for Medusa1, Medusa2 and Olympic Ridge, using the Ryzen Master SMC message set covering sensors, TjMax, overclocking enable, all core frequency, voltage ID and dynamic FCLK.

Sysadmin News

AMD Patches Bring Ryzen Master Style Controls to Linux

AMD posted kernel patches on July 29, 2026 extending the amd_hsmp driver to Zen 6 client chips, using the Ryzen Master SMC message set instead of the EPYC one.

Answer card: ChipAgents announced 60 million dollars on July 29, 2026 led by B Capital, taking its Series A to 131 million dollars across three tranches, with AI agents aimed at chip verification, the phase that consumes more than 70 percent of a design project.

Dev News

ChipAgents Raises $60M to Point Agents at Chip Verification

ChipAgents announced $60 million on July 29, 2026, led by B Capital, taking its Series A to $131 million across three tranches, with agents aimed at verification.

Answer card: the UK CMA opened an investigation on July 29, 2026 into whether Microsoft clearly told Microsoft 365 Personal and Family subscribers about the cheaper Classic plans before renewing them onto Copilot plans, a difference of 25 pounds a year on Personal.

Sysadmin News

UK Regulator Probes Microsoft 365 Copilot Price Rises

The CMA opened an investigation on July 29, 2026 into whether Microsoft gave 365 Personal and Family subscribers clear information about the Copilot price rise.

Answer card: Xen 4.22 released July 30, 2026 with per domain Xenstore quotas and watch depth limits, AMD Zen 5 Bus Lock Threshold support on x86, PSCI guest suspend to RAM and FF-A 1.2 on Arm, and the SSTC timer extension on RISC-V.

Sysadmin News

Xen 4.22 Adds Xenstore Quotas and Zen 5 Bus Lock Threshold

Xen 4.22 shipped on July 30, 2026 with per domain Xenstore quotas, AMD Zen 5 Bus Lock Threshold support on x86, PSCI suspend to RAM on Arm and FF-A 1.2.

Answer card: proposed amd_pstate epp_boost patches sample each core every 10 ms, pin the energy performance preference to performance when a core is at least 50% busy, and hold it for 300 ms, improving 1% low frame rates by 31.8% on a Steam Deck.

Sysadmin News

AMD P-State Patches Lift 1% Low Frame Rates by 31.8%

A Meta kernel engineer proposes per core EPP boosting for amd_pstate. On a Steam Deck it lifts 1% low frame rates by 31.8% in Civilization VI.

Answer card: Fish Audio raised a $52 million seed round led by Coreline Ventures and Capital Today, with $21 million ARR, more than 8 million users, five models covering 83 languages, and three of five speech generation models released as open weights.

Dev News

Fish Audio Turns 31,000 GitHub Stars Into a $52M Seed

A voice model trained on one gaming GPU became Fish Speech, then $21M of ARR across 8 million users. Three of the five models stay open, one does not.

Answer card: Meta and BlackRock will develop a 1 gigawatt data center campus in El Paso, Texas, for about $14 billion, with BlackRock funds owning 80%, Meta 20%, a $12.5 billion debt package, and Meta leasing the whole campus for up to twenty years.

Network News

Meta Keeps 20% of a 1 GW Campus and Leases Back the Rest

Meta and BlackRock will develop a $14B, 1 gigawatt campus in El Paso. BlackRock funds own 80%, a $12.5B debt package carries it, Meta leases the whole thing.

Answer card: Starling is a Linux desktop environment of 335,000 lines built in six months by one developer directing Claude, with a Wayland compositor in C, an in tree X11 server, a Swift shell, and a 273,000 line Flutter to Swift framework port.

Dev News

Starling: a Linux Desktop in Swift, Written With Claude

One developer, six months, 335,000 lines. Starling ships a Wayland compositor, an X11 server and a Swift shell, and runs Chrome and Slack unmodified.

Answer card: AMD leases about 530 MW from Core Scientific across five campuses in Texas, Oklahoma, Georgia and Alabama, on fifteen year terms worth more than $14 billion, with a reservation right on 1,925 MW more through December 2028.

Sysadmin News

AMD Leases 530 MW of Data Center Capacity for 15 Years

AMD signed fifteen year leases with Core Scientific for about 530 MW of US capacity, worth over $14 billion, plus a reservation right on nearly 2 GW more.

Answer card: the ARCTIC Fan Controller offers ten individually addressable PWM channels over USB HID, with an ARCTIC written hwmon driver mainlined for Linux 7.2, at 16.99 euros.

Sysadmin News

ARCTIC Ships a Fan Controller That Linux Already Supports

ARCTIC launched a ten channel PWM fan controller whose hwmon driver was upstreamed months before the product. It lands in Linux 7.2 and works with lm-sensors.

Answer card: GCC declines legally significant contributions that include or derive from LLM generated content, allows LLM generated test cases even when legally significant, and allows small clearly marked changes below the copyright threshold.

Dev News

GCC Will Decline LLM Written Patches, With One Exception

The GCC Steering Committee adopted a policy declining legally significant contributions derived from LLM output. Test cases are carved out, and it is provisional.

Answer card: Collabora, funded by Valve, ported the open source RADV Vulkan driver to Windows, ran Counter-Strike 2 on it, and is blocked on reverse engineered WDDM2 structures and CPU only presentation.

Dev News

Valve Pays Collabora to Port RADV to Windows

Collabora has RADV, Mesa's open source Vulkan driver for AMD GPUs, running Counter-Strike 2 on Windows. Valve funds it, and the kernel interface is the blocker.

Answer card: Moonshot AI released the full weights of Kimi K3 on Hugging Face in late July 2026, a 2.8 trillion parameter mixture of experts model with 104 billion active parameters per token, a one million token context window, MXFP4 weights totalling about 1.4 terabytes, released under the Kimi K3 License.

Dev News

Moonshot Ships Kimi K3 Weights, All 1.4 TB of Them

Moonshot AI published the full weights of its 2.8 trillion parameter Kimi K3 on Hugging Face, along with the attention kernels and MoE communication library behind it.

Answer card: Roblox began the public alpha of Build on July 28 2026 in New Zealand, a mobile tab that turns text prompts into playable games, open to age checked users nine and older with publishing restricted to sixteen and over, sharing its back end, models and chat history with Roblox Studio.

Dev News

Roblox Opens Its Prompt to Game Alpha, Judged on Retention

Build enters public alpha in New Zealand on July 28. A phone turns a prompt into a playable game, then faces the same discovery ranking as every other Roblox title.

Answer card: Canonical announced an opt in virtualization hardware enablement stack for Ubuntu 26.04 LTS in July 2026, shipping qemu-hwe, libvirt-hwe, edk2-hwe and seabios-hwe with a six month refresh cadence over the first two years of the release, managed as one unit by the ubuntu_virt_helper tool.

Sysadmin News

Ubuntu 26.04 LTS Gets an Opt In Virtualization HWE Stack

Canonical will ship qemu-hwe, libvirt-hwe, edk2-hwe and seabios-hwe on a six month cadence, so an LTS host can track newer QEMU without leaving the LTS.

Answer card: the FreeBSD Q2 2026 status report contains 41 entries, including a clean room NTSYNC driver compatible with the Linux 7.0 interface, an AMD ROCm port, GENEVE tunnel support, native MPLS from Summer of Code 2026 and full CPUID control in bhyve.

Sysadmin News

FreeBSD Wrote Its Own NTSYNC Driver, Without Linux Code

The Q2 2026 status report runs to 41 entries. The headline is a clean room NTSYNC driver matching the Linux 7.0 interface, with an AMD ROCm port in progress.

Answer card: the MCP 2026-07-28 revision removes the initialize handshake and the Mcp-Session-Id header, requires Mcp-Method and Mcp-Name headers on the Streamable HTTP transport, adds ttlMs and cacheScope to list responses, and deprecates Roots, Sampling and Logging under a twelve month policy.

Dev News

MCP Goes Stateless Today, and Sticky Sessions Go With It

The 2026-07-28 revision of the Model Context Protocol drops the initialize handshake and Mcp-Session-Id, so any request can land on any server instance.

Answer card: NVIDIA is investing a reported five billion dollars in Safe Superintelligence and supplying Vera Rubin systems, raising SSI's compute by an order of magnitude and moving the lab off Google TPUs, announced July twenty seventh 2026.

Dev News

NVIDIA Puts $5B Into Safe Superintelligence

NVIDIA takes a reported five billion dollar stake in Ilya Sutskever's lab and gives it Vera Rubin systems, moving SSI off Google TPUs and onto GPUs.

Answer card: GNU Binutils 2.47 was released on July twenty sixth 2026 by Nick Clifton, adding a linker -O 0 level that stops combining mergeable sections, --start-lib and --end-lib for loose object files, --debug-dir in objdump and readelf, and new RISC-V extensions.

Dev News

GNU Binutils 2.47 Lands With a Faster Linker Setting

Binutils 2.47 is out with an ld -O 0 level that skips mergeable section merging, --start-lib for loose object files, and a batch of new RISC-V extensions.

Answer card: in July 2026 Foxconn replaced its legacy VMware and SAN architecture with Arcfra hyperconverged software across branch factories in Mainland China, Taiwan, Vietnam and North America, covering intranet, ERP, manufacturing and production line management, DevTest and VDI workloads.

Sysadmin News

Foxconn Drops VMware for a Two Year Old HCI Vendor

Foxconn replaced VMware and SAN storage across factories in China, Taiwan, Vietnam and North America with hyperconverged software from Arcfra, a company founded in 2024.

Answer card: Paolo Bonzini merged the kvm-chainsaw branch into KVM's next tree in July 2026, splitting struct kvm_mmu into kvm_pagewalk for walking guest page tables, kvm_mmu for building shadow pages, and kvm_page_format for operating on existing PTEs, targeting Linux 7.3.

Sysadmin News

KVM Splits Its God Struct Ahead of Linux 7.3

Paolo Bonzini merged the kvm-chainsaw branch, breaking struct kvm_mmu into three. No new feature, but a large piece of nested virtualization confusion goes away.

Answer card: Shopify's successor to the Horizon storefront theme has 93 percent fewer lines of code, replacing JSON heavy configuration with HTML interpolated with Liquid, adding a composable block tag with typed parameters and a partial tag for server rendered regions, both in developer preview in July 2026.

Dev News

Shopify Rebuilt Its Theme Layer With 93% Less Code

Shopify's next storefront theme drops JSON heavy configuration for HTML and Liquid, cutting code by 93 percent. The reason given is that AI agents read HTML better.

Answer card: Linux 7.2-rc5 was released on July 26 2026 with more than 35 percent of its changes in the networking subsystem, mostly drivers, which Torvalds attributed to a conference backlog clearing, with the 7.2 stable release expected on August 16 2026.

Network News

Linux 7.2-rc5 Is Big, and a Third of It Is Networking

Linux 7.2-rc5 landed on July 26 with over 35% of its changes in networking drivers. Torvalds calls it too big but not scary. Stable targets August 16.

Answer card: NVIDIA is reported to be in talks to guarantee about 250 billion dollars of lease and construction debt so OpenAI can lease SB Energy's planned 10 gigawatt PORTS Technology Campus in Piketon, Ohio, a project whose total cost including chips could exceed 500 billion dollars.

Sysadmin News

NVIDIA Weighs a $250B Backstop for OpenAI's Ohio Campus

NVIDIA is reportedly in talks to guarantee $250 billion so OpenAI can lease SoftBank's 10 GW Piketon campus. The structure matters more than the number.

Answer card: Anthropic released Claude Opus 5 on July twenty fourth at five dollars per million input tokens and twenty five per million output tokens, half the price of Fable 5, scoring 43.3 percent on Frontier-Bench v0.1 against 33.7 percent for Fable 5.

Dev News

Claude Opus 5 Lands at Half the Price of Fable 5

Anthropic shipped Claude Opus 5 on July 24 at five dollars per million input tokens, half what Fable 5 costs, and it wins several published benchmarks.

Answer card: Debian opened the discussion period on July twenty fourth 2026 for a General Resolution on LLM usage, with four ballot options ranging from a Social Contract ban proposed by Matthias Geiger to conditional acceptance proposed by Lucas Nussbaum, Ian Jackson and Pierre-Elliott Becue.

Dev News

Debian Votes Again on LLM Written Contributions

Four competing proposals, from an outright ban to conditional acceptance, went into discussion on July 24. Whatever Debian picks, derivatives will copy it.

Answer card: glibc 2.44, released July twenty fifth 2026, introduces /etc/tunables.conf for system wide tunables as an alternative to the GLIBC_TUNABLES environment variable, adds a transparent huge pages tunable for read only executable segments, and speeds up cosh by roughly 35 percent.

Sysadmin News

glibc 2.44 Adds /etc/tunables.conf for System Wide Tuning

Released July 25, glibc 2.44 lets you set tunables in a config file instead of an environment variable, plus a THP switch and faster math.

Answer card: on July twenty second 2026 a transmission line fault in Ashburn Virginia caused roughly 3 gigawatts of data center load, about three percent of PJM demand, to transfer to backup power within seconds, taking around ten minutes to stabilise with no customer outages.

Sysadmin News

One Line Fault Dropped 3 GW of Data Center Load in Virginia

A transmission fault in Ashburn sent about 3 GW of data centers to backup power in seconds. Nobody lost power, and that is exactly what worries grid operators.

Answer card: a July twenty fourth 2026 letter titled Open Weights and American AI Leadership was signed by twenty five companies including NVIDIA, Microsoft and Meta, doubled to roughly fifty signatories within a day with OpenAI and Google joining, and asks US policymakers to avoid broad restrictions on downloadable model weights.

Dev News

Fifty Companies Sign an Open Weights Letter to Washington

NVIDIA, Microsoft, Meta and dozens more asked US policymakers not to restrict downloadable model weights. The signatory list doubled in a day and OpenAI joined.

Answer card: on July twenty fourth 2026 Ubuntu developer Benjamin Drung proposed removing /etc/debian_version from base-files, targeting Ubuntu 26.10, closing a Launchpad bug open since 2005, with /etc/os-release as the replacement.

Sysadmin News

Ubuntu Moves to Delete /etc/debian_version in 26.10

A twenty year old Launchpad bug is finally getting closed. If your scripts read /etc/debian_version to detect the distribution, now is the time to fix them.

Answer card: on July twenty third, Black Forest Labs released FLUX 3, a single multimodal model covering image, video with native audio up to twenty seconds, and robot action prediction, with FLUX 3 Video and Action in early access and an open weight FLUX 3 Dev due later in 2026.

Dev News

FLUX 3 Puts Image, Video, Audio and Robotics in One Model

Black Forest Labs launched FLUX 3, one set of weights that generates images, 20 second video with native audio, and robot actions. Open weights come later in 2026.

Answer card: on July twenty fourth, Greg Kroah-Hartman released seven stable Linux kernels at once, 7.1.5, 6.18.40, 6.12.97, 6.6.145, 6.1.178, 5.15.212 and 5.10.261, with over 2,000 patches in the 7.1.5 candidate and 1,611 in the 6.18.40 candidate.

Sysadmin News

Seven Stable Linux Kernels Shipped in One Day

Greg Kroah-Hartman released seven stable Linux kernels on July 24, from 5.10.261 up to 7.1.5, one of them carrying over 2,000 patches. Here is how to stage it.

Answer card: on July twenty third, reports said Stripe is in talks to acquire OpenRouter for close to 10 billion dollars, about eight times the 1.3 billion dollar valuation OpenRouter set in May, with no deal signed and talks that could still fall apart.

Dev News

Stripe in Talks to Buy OpenRouter for About $10 Billion

Stripe is reportedly in talks to acquire OpenRouter for close to $10 billion, roughly eight times the model routing startup's $1.3 billion valuation from May.

Answer card: on July twenty third Apple announced MapKit for Automotive, an SDK that lets carmakers embed Apple Maps as native in car navigation without CarPlay, with Ford's Universal EV Platform as the first adopter, shipping in 2027 on a vehicle priced around $30,000.

Dev News

Apple Opens MapKit for Automotive, Ford Ships It First

Apple announced MapKit for Automotive, an SDK that puts Apple Maps in the car as native navigation without CarPlay. Ford's 2027 EV platform is the first user.

Answer card: Intel Q2 2026 results announced July twenty third, total revenue $16.1 billion, data center and AI revenue $6.3 billion up 59% year over year, non-GAAP gross margin 41.8%, and 2026 capital expenditure guidance raised above $20 billion.

Sysadmin News

Intel Data Center Revenue Grows 59% on Xeon 6 Demand

Intel posted $16.1B in Q2 2026 revenue, with data center and AI up 59% to $6.3B on Xeon 6. Demand now outpaces supply, which changes how you plan a refresh.

Answer card: on July twenty fourth Qualcomm told customers that chip prices rise by a double digit percentage for shipments after September first, 2026, citing supplier costs it can no longer absorb and a shortage of alternative vendors caused by AI data center demand on component capacity.

Sysadmin News

Qualcomm Raises Chip Prices by Double Digits in September

Qualcomm told customers on July 24 that chip prices rise by a double digit percentage for shipments after September 1, blaming supplier costs it can no longer absorb.

Answer card: at Advancing AI 2026 on July twenty third, AMD launched the Instinct MI400 series GPUs, sixth generation EPYC Venice CPUs on a two nanometer process, the ROCm open software stack, and the Helios rack that combines seventy two MI455X GPUs with Venice CPUs and Pensando networking.

Network News

AMD Launches MI400 GPUs and EPYC Venice at Advancing AI

At Advancing AI 2026 on July 23, AMD launched its Instinct MI400 GPUs, EPYC Venice CPUs, and the Helios rack, with Anthropic committing to up to 2 gigawatts.

Answer card: Anthropic updated Claude voice mode on July twenty third so users can choose Opus, Sonnet, or Haiku, with the picker starting from the last model used in text chat and connections to Gmail, Google Calendar, Slack, Canva, and Notion.

Dev News

Claude Voice Mode Adds Opus and Sonnet, Not Just Haiku

Anthropic updated Claude voice mode on July 23 so you can pick Opus, Sonnet, or Haiku, with voice actions in Gmail, Calendar, Slack, Canva, and Notion.

Answer card: on July twenty third, AMD and Cerebras announced a disaggregated inference partnership, pairing AMD Helios racks for high throughput with the Cerebras Wafer-Scale Engine for fast token generation, claiming up to five times tokens per second per watt, a modelled figure, available first on Cerebras Cloud in the second half of 2026.

Sysadmin News

AMD and Cerebras Split AI Inference Across Two Chips

AMD and Cerebras will pair Helios racks with the Wafer-Scale Engine for disaggregated inference, claiming up to 5x tokens per watt, a modelled figure, in late 2026.

Answer card: on July twenty third, Etched raised 300 million dollars at a 10.3 billion dollar valuation in a Series C led by Sequoia, to scale Sohu, its transformer only inference chip, with first rack shipments due in summer 2026.

Sysadmin News

Etched Raises $300M at $10.3B to Scale Its Sohu Chip

Etched raised $300 million at a $10.3 billion valuation to scale Sohu, its transformer-only inference chip, in a round led by Sequoia with a16z and SK hynix.

Answer card: Anthropic shipped Record a Skill in Claude Cowork on July twenty first, a feature that turns a narrated screen recording of a task into a reusable Skill, available on the Pro, Max, and Team plans from the plus menu of the Claude desktop app.

Dev News

Claude Cowork Turns Screen Recordings Into Reusable Skills

Anthropic shipped Record a Skill in Claude Cowork on July 21. Screen record a task once, narrate it, and Claude turns the demo into a reusable Skill.

Answer card: Block released Buzz on July twenty first, a free Apache 2.0 open source workspace built on the Nostr protocol where humans and AI agents share channels, code repositories, and their own portable cryptographic identities.

Dev News

Block Launches Buzz, an Open Source Slack and GitHub Rival

Block released Buzz on July 21, a free Apache 2.0 open source workspace built on Nostr where humans and AI agents share channels, code, and portable identity.

Answer card: NVIDIA announced Spectrum-6 on July 21, 2026, a 102.4 Tbps Ethernet switch for the Vera Rubin platform, doubling capacity and holding 95 percent efficiency past 100,000 GPUs.

Network News

NVIDIA Spectrum-6: 102.4 Tbps Ethernet for AI Factories

NVIDIA launched Spectrum-6, a 102.4 Tbps Ethernet switch for the Vera Rubin platform that doubles capacity and holds 95 percent efficiency past 100,000 GPUs.

Answer card: OpenAI announced Project Camellia on July 22, 2026, a 3.2 gigawatt data center in Effingham County, Georgia, on a 25 year Georgia Power deal with up to 1,000 MW of demand response.

Sysadmin News

OpenAI's Project Camellia: a 3.2 GW Data Center in Georgia

OpenAI unveiled Project Camellia, a 3.2 gigawatt data center in Effingham County, Georgia, on a 25 year Georgia Power deal with up to 1,000 MW of demand response.

Answer card: OpenAI launched Presence on July 22, 2026, a limited availability platform for deploying voice and chat AI agents with policies, guardrails, simulations, and a Codex driven evaluation loop.

Dev News

OpenAI Presence Puts Enterprise AI Agents Behind Guardrails

OpenAI launched Presence, a limited availability platform for deploying voice and chat AI agents with policies, guardrails, simulations, and Codex driven evaluation.

Answer card: Google released Gemini 3.6 Flash on July twenty first, a cheaper and more token efficient Flash model for agentic coding that uses about seventeen percent fewer output tokens than 3.5 Flash and is rolling out in GitHub Copilot.

Dev News

Google Gemini 3.6 Flash Lands in Copilot, 17% Fewer Tokens

Google released Gemini 3.6 Flash on July 21, a cheaper and more token efficient Flash model tuned for agentic coding, now rolling out in GitHub Copilot.

Answer card: General Compute secured up to $400 million from Upper90 Capital Management on July 17, 2026, starting at $100 million, to build an inference neocloud on SambaNova SN40 and SN50 chips, with the loan backed by inference silicon as collateral.

Network News

General Compute Raises $400M for an Inference Cloud

General Compute secured up to $400M from Upper90 to build an inference neocloud on SambaNova SN40 and SN50 chips, using inference silicon as loan collateral.

Answer card: Kubernetes v1.37 reached code freeze on July 22, 2026, with general availability on August 26. It requires containerd 2.0 or later, makes cgroup v2 mandatory with the FailCgroupV1 gate defaulting to true, and advances Dynamic Resource Allocation and etcd RangeStream.

Sysadmin News

Kubernetes 1.37 Code Freeze: containerd 2.0 Now Required

Kubernetes v1.37 hit code freeze on July 22, 2026. It requires containerd 2.0, makes cgroup v2 mandatory and ships DRA and etcd RangeStream changes.

Answer card: on July 21, 2026, OpenAI launched a ChatGPT for small businesses program built on ChatGPT Work, an agent that acts across apps and files, can run for hours and produce finished documents, with partner skills from Dropbox, Shopify, Intuit, Slack, Atlassian and Wix.

Dev News

OpenAI Opens ChatGPT Work to Small Businesses

On July 21, 2026, OpenAI launched a ChatGPT for small businesses program built on ChatGPT Work, an agent that acts across apps and files with partner skills.

Answer card: Microsoft will deploy AMD Helios rack-scale systems on Azure, combining 72 Instinct MI455X GPUs, EPYC Venice CPUs, Pensando networking, and ROCm software.

Network News

Microsoft Will Run AMD Helios Rack Systems on Azure

Microsoft will deploy AMD Helios rack-scale systems on Azure, pairing Instinct MI455X GPUs, EPYC Venice CPUs, and Pensando networking to serve frontier AI.

Answer card: Z.AI has begun operating a 1 gigawatt data center built only on Chinese-made chips, training its GLM models on Huawei Ascend hardware and the MindSpore software stack.

Sysadmin News

Z.AI Powers Up a 1 Gigawatt Data Center on Chinese Chips

Z.AI has begun operating a 1 gigawatt data center running only Chinese-made chips, training its GLM models on Huawei Ascend hardware and the MindSpore stack.

Answer card: Alibaba previewed Qwen3.8 Max, a 2.4 trillion parameter multimodal model it ranks second only to Fable 5, with open weights promised soon.

Dev News

Alibaba Previews Qwen3.8 Max, a 2.4T Parameter Model

Alibaba previewed Qwen3.8 Max, a 2.4 trillion parameter multimodal model it ranks second only to Fable 5, with open weights promised soon.

Answer card: CXMT priced an 8.6 billion dollar Shanghai IPO, Asia largest of 2026, as the world fourth largest DRAM maker scales up server and AI memory production.

Sysadmin News

China's CXMT Prices an 8.6 Billion Dollar DRAM Memory IPO

CXMT priced an 8.6 billion dollar Shanghai IPO, Asia's largest of 2026, as the world's fourth largest DRAM maker scales up server and AI memory production.

Answer card: Hut 8 signed a second 15-year, 352 megawatt lease worth 9.8 billion dollars, fully commercializing its 1 gigawatt Beacon Point AI data center campus in Texas.

Network News

Hut 8 Fully Contracts Its 1 Gigawatt Beacon Point AI Campus

Hut 8 signed a second 15-year, 352 MW lease worth 9.8 billion dollars, fully commercializing its 1 gigawatt Beacon Point AI data center campus in Texas.

Answer card: Cloudflare shipped a wrangler flagship command suite on July 16, 2026, to create feature flags, change defaults, and flip kill switches from the terminal.

Dev News

Cloudflare Puts Feature Flags in the Terminal With Wrangler

Cloudflare's Wrangler CLI now manages Flagship feature flags from the terminal, so you can create flags, flip kill switches, and roll back from CI or a script.

Answer card: Walden Robotics launched from stealth on July 15, 2026, with about 300 million dollars at a 1.1 billion dollar valuation, co led by Deviation Capital and Toyota.

Dev News

Walden Robotics Exits Stealth With 300 Million for Robots

Walden Robotics launched from stealth with 300 million dollars at a 1.1 billion valuation, and its general purpose robots are already working on a Toyota line.

Answer card: The European Commission issued binding DMA measures on July 16 2026, ordering Google to open eleven Android features to rival AI assistants and to share anonymized search data.

Dev News

EU Orders Google to Open Android and Search to AI Rivals

The European Commission issued binding DMA measures on July 16, telling Google to open eleven Android features to rival AI assistants and share anonymized search data.

Answer card: Google renamed NotebookLM to Gemini Notebook on July 16 2026 and added a secure cloud computer that writes and runs code grounded in your own sources.

Dev News

NotebookLM Is Now Gemini Notebook, With a Cloud Computer

Google renamed NotebookLM to Gemini Notebook on July 16 and gave every notebook a secure cloud computer that writes and runs code against your own sources.

Answer card: SAP has completed its acquisition of Prior Labs, the German lab behind the TabPFN tabular foundation models, backing a European frontier AI lab with more than one billion euros over four years.

Dev News

SAP Buys Prior Labs to Lead Tabular Foundation Models

SAP has completed its acquisition of Prior Labs, the German lab behind TabPFN, backing a European frontier AI lab with more than one billion euros over four years.

Answer card: for part of Friday July 17, 2026, Apple passed Nvidia to become the world most valuable company at about 4.88 trillion dollars.

Dev News

Apple Overtakes Nvidia as World's Most Valuable Company

For part of Friday July 17, 2026, Apple passed Nvidia as the world's most valuable company near 4.9 trillion dollars. What the crown swap means.

Answer card: Japan FRONTia national AI factory will run 27,500 Nvidia Rubin GPUs at 140 megawatts to train domestic robotics foundation models.

Network News

Japan Builds World's First National AI Factory With Nvidia

Japan's FRONTia project puts 27,500 Nvidia Rubin GPUs in a 140 MW national AI factory to train robotics models. Inside the Noetra plan.

Answer card: Equinix has topped out the next phase of its MU4 data center near Munich, a site that will reach 13,608 square meters across four floors at full build out.

Network News

Equinix Tops Out Its MU4 Data Center Expansion in Munich

Equinix has topped out its MU4 expansion in Munich, adding room for thousands more cabinets at one of Germany's key interconnection metros.

Answer card: Spectro Cloud raised more than 100 million dollars in a Series D led by Goldman Sachs Alternatives to help teams run production AI on Kubernetes across clouds.

Sysadmin News

Spectro Cloud Raises 100 Million for Kubernetes AI

Spectro Cloud raised more than 100 million dollars in a Series D led by Goldman Sachs to help teams put AI on Kubernetes into production across clouds.

Answer card: Google has delayed Gemini 3.5 Pro after the model missed internal targets on coding and long horizon reasoning, pushing the flagship model months past its June launch date.

Dev News

Google Delays Gemini 3.5 Pro Over Coding and Reasoning Gaps

Google has delayed Gemini 3.5 Pro after the model missed internal targets on coding and long horizon reasoning, pushing the flagship months past its June date.

Answer card: Thinking Machines released Inkling, a 975 billion parameter open weight mixture of experts model with 41 billion active parameters, a 1 million token context, and open weights on Hugging Face.

Dev News

Thinking Machines Ships Inkling, a Leading US Open Model

Thinking Machines released Inkling, a 975 billion parameter open weight MoE with 41B active, a 1M token context, and weights on Hugging Face under an open license.

Answer card: Anthropic is in early talks to lease as much as 10 billion dollars of computing power from Meta over two years, according to a July 2026 New York Times report.

Sysadmin News

Anthropic in Talks to Lease 10 Billion in Compute From Meta

Anthropic is in early talks to lease as much as 10 billion dollars of compute from Meta over two years. What Meta Compute means for the cloud market.

Answer card: Fireworks AI closed a 1.5 billion dollar Series D at a 17.5 billion dollar valuation on July 16, 2026, led by Atreides Management, Index Ventures, and TCV.

Dev News

Fireworks Raises 1.5 Billion Series D for Inference

Fireworks AI closed a 1.5 billion dollar Series D at a 17.5 billion valuation. Why specialized open model inference is the story developers should read.

Answer card: PostgreSQL 19 Beta 2 was released on July 16, 2026, adding the REPACK command that replaces VACUUM FULL and CLUSTER, native SQL/PGQ property graph queries, JIT disabled by default and a default TOAST compression change from pglz to lz4, with general availability expected around September or October 2026.

Sysadmin News

PostgreSQL 19 Beta 2: REPACK, Graphs and New Defaults

PostgreSQL 19 Beta 2 shipped on July 16, 2026, bringing the REPACK command, SQL/PGQ property graphs, JIT off by default and lz4 TOAST compression.

Answer card: Moonshot AI released Kimi K3, a 2.8 trillion parameter open mixture of experts model with a 1M token context, thinking mode, and open weights due July 27.

Dev News

Moonshot Releases Kimi K3, the Largest Open Model Yet

Moonshot AI shipped Kimi K3, a 2.8 trillion parameter open MoE model with a 1M token context and thinking mode, with open weights due July 27.

Answer card: Anthropic bankers at Goldman Sachs, Morgan Stanley and JPMorgan are scheduling investor meetings ahead of a possible October 2026 listing, following a confidential SEC filing and a 965 billion dollar private valuation in May.

Dev News

Anthropic Lines Up Investor Meetings Ahead of an IPO

Bankers at Goldman Sachs, Morgan Stanley and JPMorgan are scheduling investor meetings for Anthropic, pointing to a listing as soon as October 2026.

Answer card: China's Cyberspace Administration approved Apple Intelligence on July 15, 2026, with Alibaba's Qwen model powering the features across iOS, iPadOS, macOS and visionOS.

Dev News

Apple Intelligence Clears China With Alibaba's Qwen

China's Cyberspace Administration approved Apple Intelligence on July 15, 2026, on the back of a deal to run Alibaba's Qwen model inside iOS, iPadOS, macOS and visionOS.

Answer card: Emergent raised a 130 million dollar Series C led by Creaegis at a 1.5 billion dollar valuation on July 15, 2026, with 120 million dollars in annual run rate revenue and over 200,000 paying customers.

Dev News

Emergent Raises 130M at 1.5B for AI App Building

Emergent, an Indian AI coding platform launched in June 2025, raised a 130 million dollar Series C at a 1.5 billion valuation with 120 million in run rate revenue.

Answer card: Apple released the iOS 27 public beta on July 13, 2026 with the redesigned AI Siri, which reaches third party apps only through App Intents after SiriKit was deprecated at WWDC on June 9, 2026, ahead of a September 14 public release.

Dev News

Apple Opens Its New AI Siri to Everyone in the iOS 27 Beta

Apple released the iOS 27 public beta on July 13, 2026, putting the redesigned AI Siri in front of everyone. For developers, App Intents is now the only way in.

Answer card: etcd v3.7.0 shipped in early July 2026 with RangeStream for chunked large query results, keys only range requests, faster lease renewal and a v3 only startup path, lowering etcd CPU use for Kubernetes clusters compared with version 3.6.

Sysadmin News

etcd 3.7 Lands: Streamed Ranges and Lower Kubernetes CPU

etcd v3.7.0 shipped in early July 2026 with RangeStream, keys only queries and a lighter startup path, cutting etcd CPU use for Kubernetes clusters versus 3.6.

Answer card: TYLsemi exited stealth on July 14, 2026 with a 43 million dollar early stage round led by Matter Venture Partners for a full stack chiplet platform, with TYL.IO and TYL.Power samples due to qualified customers in 2027 through TSMC.

Network News

TYLsemi Raises 43M to Ship Custom AI Silicon via Chiplets

TYLsemi left stealth on July 14, 2026 with a 43 million dollar round and a full stack chiplet platform for custom AI silicon, with TSMC samples due in 2027.

Answer card: New York paused state environmental permits for new data centers of 50 megawatts or more for up to one year, under an executive order signed July 14, 2026.

Sysadmin News

New York Halts New Hyperscale Data Centers for a Year

Governor Hochul signed an executive order pausing state environmental permits for new data centers of 50 megawatts or more, for up to one year.

Answer card: Reflection AI signed a compute deal worth over 1 billion dollars with Nebius, running through 2029 with access to Nvidia GB300 chips for training open weight models.

Dev News

Reflection Signs 1 Billion Dollar Nebius Compute Deal

Reflection AI signed a compute deal worth over 1 billion dollars with Nebius, running through 2029 and giving it access to Nvidia GB300 chips to train open models.

Answer card: Tower Semiconductor is investing about 3 billion dollars in 300mm silicon photonics and silicon germanium capacity in Japan, net of roughly 1 billion dollars in government grants.

Network News

Tower Puts 3 Billion Dollars Into Japan for Optical Chips

Tower Semiconductor is investing about 3 billion dollars in Japan for 300mm silicon photonics and silicon germanium, backed by roughly 1 billion in grants.

Answer card: ASML reported 9.3 billion euros of net sales in the second quarter of 2026 and raised its full year outlook to between 43 and 45 billion euros.

Sysadmin News

ASML Lifts 2026 Outlook as Intel Ships High NA EUV Chips

ASML reported 9.3 billion euros of Q2 net sales and raised its 2026 outlook to 43 to 45 billion euros, as Intel became the first to ship logic made with High NA EUV.

Answer card: Linux 7.2 adds cache aware scheduling, enabled with CONFIG_SCHED_CACHE and toggled at runtime through the llc_balancing debugfs knob.

Sysadmin News

Linux 7.2 Adds Cache Aware Scheduling for Big Server CPUs

Linux 7.2 reached rc3 on July 12, 2026, carrying cache aware scheduling, an Intel led feature that keeps tasks sharing data on the same last level cache.

Answer card: Rakuten Mobile completed its nationwide deployment of 1Finity Open RAN radio units in May 2026 and plans around 3,000 massive MIMO radios at 3.7 GHz.

Network News

Rakuten Mobile Finishes Nationwide Open RAN Radio Rollout

Rakuten Mobile has deployed 1Finity Open RAN radios across its nationwide 5G network in Japan, with a further 3,000 massive MIMO units planned at 3.7 GHz.

Answer card: Intel is investing 5 billion euros in its Leixlip campus in Ireland to scale Xeon 6 and next generation Xeon on the Intel 3 node, after regaining full control of Fab 34.

Sysadmin News

Intel Puts 5 Billion Euros Into Ireland for More Xeon

Intel is investing 5 billion euros in its Leixlip campus in Ireland to scale Xeon 6 and next gen Xeon on the Intel 3 node, after buying back full control of Fab 34.

Answer card: Meta will move its fourth generation MTIA chip, Iris, into production in September 2026, designed with Broadcom and built by TSMC, to double compute and reduce reliance on Nvidia and AMD.

Sysadmin News

Meta Moves Its Iris AI Chip Into Production in September

An internal memo says Meta will put its fourth generation MTIA chip, Iris, into production in September 2026 to double compute and lean less on Nvidia and AMD.

Answer card: OpenAI shipped the GPT-5.6 family, Sol, Terra, and Luna, alongside ChatGPT Work, a desktop agent with Codex built in that turns a goal into finished work.

Dev News

OpenAI Ships GPT-5.6 and ChatGPT Work, a Desktop Agent

OpenAI released the GPT-5.6 family (Sol, Terra, Luna) plus ChatGPT Work, a desktop agent with Codex built in that turns a goal into finished docs, sheets, and code.

Answer card: Cloudflare Monetization Gateway uses the x402 protocol and the HTTP 402 status code to let sites charge AI agents per request for pages, APIs, datasets and MCP tools, settled in stablecoins.

Network News

Cloudflare Charges AI Agents Per Request via HTTP 402

Cloudflare's new Monetization Gateway uses the x402 protocol and HTTP 402 to let any site charge callers per request for pages, APIs, datasets and MCP tools.

Answer card: Mistral Leanstral 1.5 is an open weights Apache 2.0 model for Lean 4 formal verification that solved 587 of 672 PutnamBench problems and found 5 unknown bugs across 57 real repositories.

Dev News

Mistral Open Sources Leanstral 1.5 for Lean 4 Proofs

Mistral released Leanstral 1.5, an open weights Apache 2.0 model for Lean 4 that proves code correct and found 5 unknown bugs across 57 real repositories.

Answer card: Meta will spend about 10 billion dollars on a Sturgeon County Alberta data center drawing near 1 gigawatt, scalable to 1.8 gigawatts, its 33rd worldwide.

Sysadmin News

Meta Bets 10 Billion Dollars on a 1 GW Alberta Data Center

Meta will spend about 10 billion dollars on its first Canadian data center in Alberta, drawing near 1 gigawatt. Why power, not chips, now sets the pace.

Answer card: SambaNova closed the first part of a 1 billion dollar Series F at an 11 billion dollar valuation led by General Atlantic to scale on premises AI inference.

Dev News

SambaNova Raises 1B at 11B to Challenge Nvidia on Inference

SambaNova closed the first part of a 1 billion dollar Series F at an 11 billion valuation to scale on premises AI inference. Why the anti Nvidia bet grew.

Answer card: SK Hynix raised 26.5 billion dollars in the biggest foreign US IPO ever, pricing 177.9 million ADRs at 149 dollars to fund HBM and fab expansion.

Sysadmin News

SK Hynix 26.5B Nasdaq Debut Bets Big on AI Memory

SK Hynix raised 26.5 billion dollars on Nasdaq, the biggest foreign US IPO ever, to fund HBM and fab expansion. What it means for AI memory supply.

Answer card: Qwen 3.7 open weights are not out yet, so run the newest open Qwen 3.6 with Ollama today using ollama run qwen3.6, pick a size for your VRAM, and swap the tag to qwen3.7 when the open weights arrive.

Dev Guide

How to Run Qwen 3.7 Locally with Ollama (2026)

How to run Qwen locally in 2026: Qwen 3.7 open weights are not out yet, so we set up the newest open Qwen with Ollama, by size, with a one-tag swap for 3.7.

Answer card for the awk cheatsheet: print columns with $1 and $NF, change the field separator with -F, filter rows by pattern, sum a column, and run BEGIN and END blocks.

Dev Guide

The awk Cheatsheet: Process Text in Columns

A practical awk cheatsheet: print columns with $1 and $NF, set field separators with -F, filter rows by pattern, sum a column, and use BEGIN and END blocks.

Answer card for the dig and nslookup cheatsheet: query A and AAAA records, MX, TXT, NS and CNAME, trim with +short, trace from the root with +trace, ask @8.8.8.8, and reverse lookup with dig -x.

Network Guide

dig and nslookup Cheatsheet: Query DNS Fast

A practical dig and nslookup cheatsheet: look up A, AAAA, MX, TXT, NS and CNAME records, trim noise with +short, trace from the root with +trace, and query @8.8.8.8.

Answer card for the everyday Docker commands: run a container, list with ps, read logs, exec into a shell, bring a stack up with compose, and prune to reclaim disk.

Sysadmin Guide

Docker Commands Cheatsheet: The Everyday Ones

A practical Docker cheatsheet: run a container with docker run, list with ps, follow logs, exec into a shell, bring a stack up with compose, and prune disk.

Answer card for the everyday git cheatsheet: check status, stage and commit, branch and switch, and undo work with restore, reset, revert, stash and reflog.

Dev Guide

The Git Cheatsheet: Everyday Commands and Quick Fixes

A practical git cheatsheet: check status, stage and commit, switch branches, and get out of trouble with reset, restore, revert, stash and reflog.

Answer card for the sed command cheatsheet: substitute with s slash old slash new slash g, edit in place with -i and a backup, delete lines with d, and print specific lines with -n p.

Dev Guide

The sed Command Cheatsheet: Substitute and Delete

A practical sed cheatsheet: substitute with s/old/new/g, edit files in place with -i and a backup, delete lines, and print specific lines with -n p.

Answer card: Anthropic disabled Claude Fable 5 and Mythos 5 worldwide on June 12 2026 after a US government export-control order, with Opus 4.8 the live fallback.

Dev News

Anthropic Suspends Claude Fable 5 and Mythos 5

Anthropic disabled Claude Fable 5 and Mythos 5 worldwide after a US government order. The timeline, the trigger, and how to fall back to Opus 4.8 fast.

Answer card: AI in cybersecurity in 2026 is a dual-use force multiplier, the same capability helping defenders triage and attackers exploit.

Security

AI in Cybersecurity 2026: Real vs Hype, Both Sides

AI in cybersecurity in 2026: what it really does for defenders (triage, summarization, code review), what attackers do with it, and the dual-use problem.

Answer card: most of the bug bounty skill you need is sysadmin DNA already, and the report is what gets paid.

Security

Bug Bounty Starter Guide for Sysadmins

A bug bounty starter guide for sysadmins: how your DNS, HTTP and Linux skills transfer, the five free tools I open, and the report shape that gets paid.

Answer card: run a capable local AI agent with Ollama and Qwen 2.5, install on Windows, Linux or macOS, pick the model size by VRAM, call it from Python and LangChain.

Dev Guide

Build a Local AI Agent with Ollama + Qwen 2.5

How I run a local AI agent with Ollama and Qwen 2.5 in 2026: install on Windows, Linux or macOS, pick the model your VRAM fits, call it from Python and LangChain.

Answer card for migrating off end-of-life CentOS 7 or 8 to Rocky Linux 9 or AlmaLinux 9: two routes, in-place ELevate or fresh install, with a mandatory pre-flight inventory.

Sysadmin Guide

CentOS to Rocky Linux 9 Migration Playbook 2026

CentOS to Rocky Linux 9 migration, end to end: the in-place ELevate route, the fresh-install route, the pre-flight inventory, and the breaks that fail silently.

Answer card comparing ChatGPT, Claude and Gemini in 2026, with no single winner because the best model depends on the job.

Dev

ChatGPT vs Claude vs Gemini 2026

A hands-on ChatGPT vs Claude vs Gemini comparison for 2026, ten real jobs tested through the APIs: code, research, writing, data, vision, cost and latency.

Answer card for the chmod and chown cheatsheet: chmod +x as the Permission denied fix, 755 and 644 as the everyday modes, chmod 600 for secrets, chown -R for ownership, and read 4 write 2 execute 1.

Sysadmin Guide

The chmod and chown Cheatsheet, Decoded

A practical Linux chmod and chown cheatsheet: the Permission denied fix with chmod +x, why 755 and 644 keep showing up, chown for ownership, and recursing safely.

Answer card comparing Claude Fable 5 and Opus 4.8: new Mythos-class tier above Opus, 95.0% vs 88.6% on SWE-bench Verified, exactly 2x the price.

Dev

Claude Fable 5 vs Opus 4.8: One Day In

Claude Fable 5 vs Opus 4.8 after a mid-session swap on launch day: 95.0% vs 88.6% SWE-bench, exactly 2x the price, and the safety routing caveat.

Answer card for choosing a Linux distribution in 2026: weigh team skills, compliance, app stack and support model in order, then fall through to Ubuntu 24.04 LTS.

Sysadmin

Choosing a Linux Distribution in 2026: Decision Guide

Choosing a Linux distribution in 2026 without the tribal noise: four inputs that decide (team skills, compliance, app stack, support) plus honest distro profiles.

Answer card for Core Web Vitals 2026: LCP under 2.5 seconds, INP under 200 milliseconds and CLS under 0.1, all judged at the 75th percentile over a rolling 28 days.

SEO

Core Web Vitals 2026: LCP, INP and CLS Mastery

Master Core Web Vitals in 2026: what LCP, INP and CLS really measure, lab vs field data, the fixes that still hold up, plus budgets and a weekly workflow.

Answer card for building a CTF lab on Hyper-V: three trust-segmented virtual switches feeding a Kali attacker, pfSense, Metasploitable3, a broken Windows 10 and DVWA.

Sysadmin Guide

Build a CTF Lab on Hyper-V: Full Walkthrough

Build a CTF lab on Hyper-V: three trust-segmented virtual switches, a Kali attacker, Metasploitable3 and a broken Windows 10, with pfSense in the middle for routing.

Answer card for the curl cheatsheet: download and keep the remote name with curl -O, follow redirects with -L, fetch headers only with -I, and POST JSON with -d.

Dev Guide

curl Cheatsheet: Download Files and Test APIs

A practical curl cheatsheet: download and keep the remote name with curl -O, follow redirects with -L, fetch headers only with -I, and POST JSON with -d.

Answer card for DNS over HTTPS: lookups wrapped in TLS 1.3 on port 443, native on Windows, Linux and macOS, self-hosted with unbound plus dnsdist.

Network Guide

DNS over HTTPS Implementation Guide

DNS over HTTPS the way I roll it out: what DoH buys you (and what it does not), flipping it on for Windows, Linux and macOS, plus self-hosting and edge enforcement.

Answer card: an ESXi host Not Responding in vCenter is a dead management link, not a dead host. Restart hostd then vpxa, free a full log partition, check ports 902 and 443, then reconnect.

Sysadmin Guide

Fix: ESXi Host Not Responding in vCenter

ESXi host Not Responding in vCenter is almost always a dead management link, not a dead host. Restart hostd then vpxa, clear the log partition, check ports 902/443.

Answer card for the Linux find command cheatsheet: find by name, by type, by size, by time, delete matches, and run a command on each result with -exec.

Dev Guide

The find Command Cheatsheet: Recipes You Actually Use

A practical Linux find cheatsheet: find by -name, big files with -size +100M, recent ones with -mtime, delete with -delete, and run -exec on each hit.

Answer card comparing git rebase and merge: merge preserves history with a two-parent commit, rebase rewrites it into a straight line.

Dev Guide

Git Rebase vs Merge: A Visual Workflow Guide

Git rebase vs merge, decided in one mental model and five real situations: branch diagrams before and after, the exact commands, and the rule I never break.

Answer card for the Linux grep cheatsheet: search one file with grep pattern file, search a tree with grep -rn, ignore case with -i, invert with -v, and choose -E or -F.

Dev Guide

The grep Cheatsheet: Search a File, Search a Tree

A practical Linux grep cheatsheet: search one file with grep pattern file, search a whole tree with grep -rn, ignore case with -i, invert with -v, and pick -E or -F.

Answer card for fixing 'the hypervisor is not running' in Hyper-V: five gates in order, BIOS virtualization, the feature, hypervisorlaunchtype, conflicts, then services.

Sysadmin Guide

Hyper-V: The Hypervisor Is Not Running, Fix Guide

Fix Hyper-V 'the hypervisor is not running' by working the five gates in order: BIOS virtualization, the feature, hypervisorlaunchtype, conflicts, services.

Answer card for image optimization in 2026: serve AVIF first through picture, a four-width responsive srcset, lazy-load everything but the hero, compress to the lowest quality you cannot see, and pin width and height.

SEO Guide

How I Optimize Images for Website Speed in 2026

Image optimization techniques for 2026: why AVIF goes first with WebP and JPEG behind it, responsive srcset, the lazy-load plus LCP trap, compression, and CDN automation.

Answer card: keep INP under 200 milliseconds at the 75th percentile, the good threshold for the Core Web Vital that replaced FID.

SEO

INP Optimization 2026: Fix Interaction to Next Paint

A hands-on INP optimization guide for 2026: the twelve Interaction to Next Paint killers I keep finding in audits, how to profile them, and a real 480ms to 120ms fix.

Answer card mapping iptables to nftables: list rules, allow SSH, default drop, persist, and translate.

Network Guide

iptables vs nftables: A Linux Firewall Cheatsheet

iptables vs nftables, side by side: list rules, allow SSH on dport 22, default drop policy, persist to rules.v4 or nftables.conf, convert with iptables-translate.

Answer card for the Linux firewall comparison: iptables, nftables, firewalld and ufw all write rules to one nf_tables kernel backend in 2026.

Network

Linux Firewall Comparison 2026: ufw, nftables, iptables

Linux firewall comparison for 2026: iptables, nftables, firewalld and ufw all poke one kernel. Per-distro defaults, the same rule four ways, plus my honest pick.

Answer card: 32 sysctl hardening toggles grouped into network, kernel, filesystem and process plus BPF plus modules, sorted by criticality.

Security Guide

Linux Kernel Hardening: sysctl Checklist

Linux kernel hardening with sysctl: the 32 toggles I actually ship, grouped by subsystem and sorted by how much each one matters, plus a drift check.

Answer card mapping deprecated net-tools commands to their iproute2 replacements: ifconfig to ip a, netstat -tulpn to ss -tulpn, route to ip route, arp to ip neigh.

Network Guide

Linux Networking Commands: ip and ss Cheatsheet

ifconfig and netstat are gone on most modern Linux installs. The iproute2 replacements you need: ip a, ss -tulpn, ip route, ip neigh, and a clean old-to-new map.

Answer card comparing a matched MikroTik RouterOS stack against a Ubiquiti UniFi stack for a prosumer home network, both around 400 EUR.

Network

MikroTik vs Ubiquiti for Home Networks (2026)

MikroTik vs Ubiquiti for a prosumer home network: matched stacks under 400 EUR, CLI vs UI, real routing protocols, Wi-Fi roaming, VPN, a 3-year cost breakdown.

Answer card for mobile optimization in 2026: mobile is the default Google ranks, the three Core Web Vitals targets (LCP under 2.5s, INP under 200ms, CLS under 0.1) at the 75th percentile, and the usual culprits.

SEO Guide

Mobile Optimization 2026: Fast and Findable on Phones

Mobile optimization in 2026: the Core Web Vitals targets for mobile (LCP under 2.5s, INP under 200ms, CLS under 0.1), what slows a phone, and how to test like one.

Answer card: five Linux network stacks in 2026, netplan, ifupdown, NetworkManager, systemd-networkd and wicked, doing the same static IP job across seven distros.

Network

Network Config Across Linux Distros in 2026

Network config across Linux distros in 2026: netplan, ifupdown, NetworkManager, systemd-networkd and wicked compared, with the static IP recipe for each.

Answer card for the nmcli cheatsheet: scan Wi-Fi with nmcli device wifi list, join with nmcli device wifi connect SSID password, show saved profiles, and set a static IP with ipv4.method manual.

Network Guide

nmcli Cheatsheet: Manage Wi-Fi From the Terminal

A practical nmcli cheatsheet for NetworkManager: scan and connect to Wi-Fi, list saved connections, bring them up and down, check device status, and set a static IP.

Answer card comparing pfSense CE 2.7 and OPNsense 24.7 for a home network: same FreeBSD core and protocols, different release cadence, UI, and plugin ecosystem.

Network

pfSense vs OPNsense for Home Networks (2026)

pfSense vs OPNsense for home networks, run side by side on one N100 mini-PC for years: UI, update cadence, plugins, real benchmarks, appliances, and which I pick.

Answer card for the PowerShell networking cheatsheet: Test-NetConnection with -Port to check a port, Get-NetIPConfiguration for IP and DNS, Get-NetTCPConnection for listening ports, Resolve-DnsName for lookups.

Network Guide

PowerShell Networking Cheatsheet for Windows

A PowerShell networking cheatsheet: test a port with Test-NetConnection, read your IP and DNS with Get-NetIPConfiguration, find listening ports, resolve names.

Answer card comparing Proxmox VE, VMware ESXi and Hyper-V in 2026 across cost, clustering, backup and containers, with Proxmox the default pick for most.

Sysadmin

Proxmox vs ESXi vs Hyper-V in 2026: Which to Run

Proxmox vs ESXi vs Hyper-V in 2026: honest cost, licensing, clustering, backup and containers after the Broadcom shakeup, plus which fits homelab or enterprise.

Answer card: a Python regex cheatsheet of 20 tested re patterns for email, IPv4, URL, UUID, log lines and more, each with its edge case.

Dev Guide

Python regex cheatsheet: 20 essential patterns explained

A Python regex cheatsheet of 20 tested re patterns: email, IPv4 and IPv6, URL, phone, ISO 8601, UUID, CSV split, log lines, slugs, with the edge case for each.

Answer card for the rsync cheatsheet: mirror a folder with rsync -av, copy over ssh with rsync -avz, mirror exactly with --delete, preview with a dry run, resume with -avP, and the trailing slash gotcha.

Sysadmin Guide

The rsync Cheatsheet: Mirror, Sync and Copy Over SSH

A practical Linux rsync cheatsheet: mirror with rsync -av, copy over ssh with -avz, mirror exactly with --delete, dry-run, resume, and the trailing slash gotcha.

Answer card for hardening an Ubuntu 24.04 server: keys-only SSH on a custom port, UFW default deny, fail2ban, unattended-upgrades, AppArmor enforce, auditd.

Security Guide

Ubuntu 24.04 Server Hardening: SysAdmin Checklist

Ubuntu 24.04 server hardening, the way I run it on every VPS: keys-only SSH on a quiet port, UFW, fail2ban, unattended updates, AppArmor enforcing, auditd.

Answer card for a self-hosted WireGuard VPN: server on a 5 EUR VPS, split vs full tunnel, Curve25519 keys, NAT and DNS done right.

Network Guide

Self-Hosted VPN with WireGuard: Setup Guide

Self-hosted WireGuard VPN on a 5 EUR VPS: pick split vs full tunnel, generate Curve25519 keys, write the server and client configs, and get NAT and DNS right.

Answer card for a SOC homelab: Wazuh host monitoring and SIEM, Suricata network IDS, and the Elastic Stack for storage and search, on one small VM.

Security Guide

SOC Homelab with Wazuh, Suricata and ELK

Build a SOC homelab with Wazuh, Suricata and the Elastic Stack on a spare VM: the single-node installer, agent enrolment, and test attacks you can fire yourself.

Answer card for running Nagios Core on a Hyper-V Ubuntu 24.04 VM: build from source, probe HTTP, SSH, MySQL and disk, alert by mail and Telegram.

Sysadmin Guide

Nagios on Hyper-V: Server Monitoring Setup Guide

Server monitoring with Nagios on a Hyper-V Ubuntu 24.04 VM: build Nagios 4.5 from source, aim HTTP, SSH, MySQL and disk probes, then wire mail and Telegram alerts.

Answer card for the SSH command cheatsheet: connect with ssh user@host, make an ed25519 key, push it with ssh-copy-id, copy files with scp, and tunnel with -L and -D.

Sysadmin Guide

SSH Command Cheatsheet: Connect, Keys, scp, Tunnels

A practical SSH command cheatsheet, most-used first: connect with ssh user@host, make an ed25519 key, push it with ssh-copy-id, copy files with scp, tunnel ports.

Answer card for SSH Permission denied (publickey): read the handshake, then close the first failing gate, from agent to authorized_keys to permissions to sshd to SELinux.

Sysadmin Guide

SSH Permission Denied (publickey): The Fix

Permission denied (publickey) is never mysterious once you read the handshake. The real causes in order, each with the exact command to confirm and fix it.

Answer card for the systemctl and journalctl cheatsheet: status to check a service, restart and reload, enable --now to start now and on boot, daemon-reload after a unit edit, and journalctl -u NAME -f to tail logs live.

Sysadmin Guide

systemctl and journalctl Cheatsheet

A systemctl and journalctl cheatsheet: status to see if a service runs and why it died, restart, reload, enable --now, daemon-reload, and journalctl -u NAME -f.

Answer card: Tailscale is WireGuard plus a control plane that distributes keys, traverses NAT via DERP, runs MagicDNS and enforces ACLs.

Network

Tailscale vs WireGuard in 2026: Not Competitors

Tailscale vs WireGuard in 2026: Tailscale is WireGuard plus a control plane for keys, NAT traversal via DERP, MagicDNS and ACLs. When each one wins.

Answer card for the Linux tar command cheatsheet: create a gzip archive with tar -czf, extract with tar -xzf using the eXtract Ze File mnemonic, list with tar -tzf, and pick a compressor.

Sysadmin Guide

The tar Command Cheatsheet: Create, Extract, List

A practical Linux tar cheatsheet: create a gzip archive with tar -czf, extract with tar -xzf (eXtract Ze File), list with tar -tzf, plus exclude and strip-components.

Answer card for Core Web Vitals 2026: LCP under 2.5 s, INP under 200 ms, CLS under 0.1, all judged at the 75th percentile of real field data.

SEO Guide

Core Web Vitals 2026: Fix LCP, INP and CLS (Real Code)

Core Web Vitals in 2026: before/after code I shipped for LCP, INP and CLS, reading a Lighthouse capture honestly, plus the WordPress and Cloudflare toggles I reach for.

Answer card for self-hosting Vaultwarden: one Docker container behind Caddy with automatic TLS, first admin signup, signups disabled, then fail2ban, restic backups and a locked admin route.

Security Guide

Vaultwarden Self-Host Guide: Bitwarden in 20 Minutes

Vaultwarden self-host guide: docker compose plus Caddy with automatic TLS, your first admin signup, then fail2ban, restic backups, and locking down the admin route.

Answer card for a VMware ESXi PSOD: read the build, exception type, top backtrace frame and coredump status, then recover the host.

Sysadmin Guide

VMware ESXi PSOD: Diagnose and Recover the Host

VMware ESXi PSOD fix guide: read the purple screen, tell a bad driver from dying hardware, grab the coredump before you reboot, then recover the host.

Answer card: two ways to run VMware Workstation and Hyper-V together, keep Hyper-V and upgrade VMware to 16+ on the Windows Hypervisor Platform, or disable Hyper-V and VBS for legacy VMware 15.

Sysadmin Guide

VMware Workstation and Hyper-V on the Same PC (2026)

The VMware Device/Credential Guard not compatible error has two fixes: keep Hyper-V and upgrade VMware to Workstation 16+, or disable Hyper-V and VBS for VMware 15.

Answer card for Wazuh SIEM on a budget: a Raspberry Pi 5 under 100 EUR, an Intel N100 mini-PC, or a Hetzner CAX21 ARM cloud node under 10 EUR a month, all running the single-node bundle.

Security Guide

Wazuh SIEM on a Budget: Pi, N100 or Cloud

Wazuh SIEM on a budget, three paths I actually run: a Raspberry Pi 5 under 100 EUR, an Intel N100 mini-PC, or a Hetzner CAX21 ARM node under 10 EUR a month.

Answer card: target WCAG 2.2 level AA, kill the five issues that sink most sites, and test with an automated pass, a keyboard pass and a screen-reader pass.

SEO Guide

Web Accessibility and WCAG in 2026

Web accessibility in 2026: target WCAG 2.2 AA, fix the five issues that sink most sites (contrast, alt text, labels, link text, page language), and test it yourself.

Answer card for Web Workers in 2026: offload heavy JavaScript to another thread so the main thread stays free and INP drops under 200 ms.

Dev Guide

Web Workers in 2026: Patterns, Pitfalls and INP

Web Workers in 2026: the patterns I reach for, the ones I skip, transferables, SharedArrayBuffer, Comlink, and how to prove the move actually helped your INP.

Answer card for a 2026 web app security audit worked in four layers: transport, application code, supply chain and secrets, each mapped to OWASP 2025.

Security Guide

Web App Security Audit 2026: A Practical Method

The web app security audit I actually run in 2026, in four layers: transport, app code, supply chain, secrets, with the tools and a 25-point checklist.

Answer card for a 10-step WordPress security audit in 2026: PHP and WP versions, plugin CVEs, HTTP headers, htaccess, TLS, login 2FA, xmlrpc, user enumeration, logging, off-site backups.

Security Guide

WordPress Security Audit in 10 Steps (2026)

A WordPress security audit you run in about 90 minutes: PHP and WP versions, plugin CVEs, HTTP headers, htaccess, TLS, 2FA, xmlrpc, user enumeration, logs and backups.