AI Compute & Chips GKE Dynamic Slicing Assembles TPU Slices When a Workload Asks for One "GKE dynamic slicing builds TPU slices at scheduling time from pre-provisioned 16-VM cubes, and Google reports job startup up to 5x faster with recovery 4.5x faster than building a static no
AI Compute & Chips NVIDIA Opens Its Scale-Up Fabric to Korean NPUs NVIDIA, Rebellions and SK Telecom will connect Korean R200 NPUs to NVLink Fusion and the MGX rack, split inference between GPU prefill and NPU decode, and aim for deployment in 2028.
AI Compute & Chips NVIDIA Answers HBM Pressure With Tiering: 3% More Memory, 30x More Throughput per Dollar "NVIDIA's 5 November memory post pairs 3% more HBM per GPU with up to 30x higher throughput per dollar, and puts the rest of the gain in NVFP4 weights, KV cache offload to CMX storage and RA
AI Compute & Chips Three MoE Inference Optimizations NVIDIA Measured on DGX Spark NVIDIA took a 124-billion-parameter mixture-of-experts model from 27.71 ms to 9.72 ms per output token on a DGX Spark, then re-ran it on 1,319 GSM8K questions to find which part of that gain
AI Compute & Chips NVIDIA Mission Control 2.5 Brings Autonomous Recovery to Kubernetes Jobs Mission Control 2.5 extends NVIDIA's autonomous recovery engine from Slurm to Kubernetes workloads scheduled by Run:ai, with job retries running in parallel to node diagnostics.
AI Compute & Chips CUDA 13.4 Turns a GPU Fraction Into an Enforceable Memory Limit NVIDIA shipped CUDA 13.4 with an application-transparent memory boundary and an open-source GPU Fractioning Operator that lets any Kubernetes scheduler enforce per-workload shares of one GPU
AI Compute & Chips Vertiv Sells a Rack-Scale AI Package Validated Before It Ships Vertiv's SmartIT arrives as one order covering rack, DC power conversion, direct-to-chip cooling and monitoring, validated at the factory, and the work it leaves behind sits with the buildin
AI Compute & Chips SGLang Runs DeepSeek's Own Benchmarks Against DeepInfer and Matches It The SGLang team measured its engine against DeepSeek's freshly open-sourced DeepInfer on DeepInfer's own benchmarks, reaching up to 1.04x prefill throughput and shorter decode steps on the s
AI Compute & Chips NVIDIA Launches Minerva and Measures Observability in Factory Yield NVIDIA launched Minerva on 22 October 2026, an AIOps platform that reads telemetry from GPUs, Spectrum-X switches, SuperNICs, DPUs, Slurm and Run:ai, and reports on AI factory yield measured
AI Compute & Chips Gartner Sizes the AI Capacity Bill at $8.5 Trillion and Counts $3.6 Trillion of Revenue Gartner's 14 October 2026 analysis says AI data centre owners would need $8.5 trillion of cumulative revenue by 2030 against no more than $3.6 trillion attributable to AI, exposing over $3.5
AI Compute & Chips AMD Helion Sizes a Rack Around the Work Agents Do Between Model Calls AMD introduced Helion on 12 October 2026 as an OCP rack blueprint for agentic AI, with a CPU-only shape reaching 32,768 EPYC cores, an all-in-one shape combining EPYC nodes with Instinct MI3
AI Compute & Chips Microsoft's Maia 300 Publishes Rack Numbers for Its Own Inference Silicon Microsoft announced Maia 300 on 12 October 2026 and framed it as a complete rack system. The chip carries eight HBM4 stacks inside a 950 W package, the rack holds 96 accelerators in a 167-ki
AI Compute & Chips SparseEngine Puts Fifteen Sparse Attention Methods Behind One Lifecycle Contract A research team published SparseEngine, an inference engine designed around sparse attention from the start. It runs 15 sparse methods across four families on a shared
AI Compute & Chips Scality's AI Inference Factory Keeps the Context Cache in Object Storage Over RDMA Scality shipped a validated stack for running open-weight models on customer-owned servers, and its storage layer holds the context cache that normally lives inside GPU memory.
AI Compute & Chips AMD's Trillion-Dollar Mark Rests on the CPU Inside Personal Agents AMD closed at $649.42 on 6 October 2026 and passed a trillion dollars in market value while both flagship personal agents disclose the CPU underneath them: two cores per Muse sandbox, nine p
AI Compute & Chips The Rack Is Now a Power Contract: Density Numbers Decide Where Compute Lands Rack density moved from about 16 kW to 27 kW in a single year, and the newest AI racks are specified up to 246 kW. That makes the electricity room, the fluid loop and the grid queue the deci
AI Compute & Chips DGX Spark at 64GB: What $4,999 of Local Compute Buys a Small Team NVIDIA's 64GB DGX Spark configuration lands on 23 October from $4,999, runs models up to 100 billion parameters on the desk, and clusters two units into 128GB of pooled memory. The interesti
AI Compute & Chips Anthropic's $11.6B Akamai Commitment: The Commercial Economics of the CPU Serving Layer Akamai's $11.6 billion, seven-year commitment with Anthropic on September 24, 2026 is for CPU workloads — serving, orchestration, agents, delivery — not GPU training. The warrant ties Anthro
AI Compute & Chips NVIDIA Topograph Treats GPU Placement as a Live Topology Problem NVIDIA's open-source Topograph discovers cluster topology from cloud APIs or on-prem fabrics and publishes it as Kubernetes labels, Slurm config, or Slinky ConfigMaps. Netics on why topology
AI Compute & Chips Benchmarking LLM Inference Needs a Real Load Client NVIDIA's AIPerf replaces GenAI-Perf with a multiprocess load client that stops being the bottleneck. Netics on why LLM inference numbers are only worth as much as the client generating them.
AI Compute & Chips Quantum Readiness Gets a Scoreboard: CUDA-Q Logical and the QUOPS Benchmark NVIDIA's CUDA-Q Logical orchestration layer and Sandia's QUOPS benchmark give fault-tolerant quantum computing something it lacked: a measurable readiness test. Netics explains why that matt
AI Compute & Chips CUDA Rust's Two Tracks Put Memory Safety Back in the Compiler NVIDIA ships two compiler tracks for writing GPU kernels in Rust. Netics explains why compile-time memory safety changes kernel engineering more than the language switch does.
AI Compute & Chips NVIDIA’s IBC 2026 Media Stack Puts Real-Time AI on Trial NVIDIA’s IBC 2026 announcements connect live inference, media exchange and localization. Netics maps the controls broadcasters should validate before deployment.
AI Compute & Chips Google’s Finland AI Investment Is a Regional Infrastructure Strategy Google’s €13 billion Finland plan links data centers, power, heat reuse, skills, and local suppliers. Its real test is regional coordination.
AI Compute & Chips NVIDIA’s IFA 2026 Push Makes Local Agents a Systems Problem NVIDIA’s IFA 2026 announcements make local agents easier to run, but hardware access, evaluation, routing, security and lifecycle discipline still decide whether they work.