Fetching from the wire…
01
02
03
04
05
06
07
08
Policy2026-08-06 · source-backed
Announced August 5, funding up to 10 hubs, one award per state or region, pooling advanced computing, data and software for AI-enabled research and education, with NVIDIA, AMD, Intel, Dell, Hangar and the Secunda Innovation Fund all committing support. It also funds consortium coordination, AI-infrastructure workforce training, and curriculum development. A public-compute counterweight at a moment when frontier capacity is concentrating in a handful of private clouds. $100M against SpaceX's $18.4B quarterly capex is a rounding error, but it's a rounding error pointed at universities.
Each link below shares sources, entities, or timing with this story.
PR #19378 landed in llama.cpp this week, and I think most people are underselling what it means. Backend-agnostic tensor parallelism via --split-mode tensor makes multi-GPU inference work across AMD, Intel, and Apple Silicon. Not just CUDA. Everything. For context: llama.cpp h...
The compute starts serving Anthropic traffic in late 2027 (TechCrunch). It's the largest in a run of deals over eight months: $10B with Volta in August, $5B with AMD in July, a reported $1.25B monthly from SpaceX in May, 5 additional gigawatts from Amazon in April. The lead ti...
SpaceX's S-1 surfaced that rival Anthropic rents 300MW (~220,000 GPUs) of xAI's Colossus 1 capacity near Memphis for $1.25B a month through May 2029, a contract that could exceed $40B, terminable on 90 days' notice. xAI posted a ~$2.4B Q1 loss against $7.7B capex. Competitors...
AMD unveiled its first rack-scale system to directly contest Nvidia at the rack level, with engineering samples in H2 2026 and mass production targeted Q2 2027. Microsoft joins Meta, OpenAI and Oracle as customers; Meta plans 1 gigawatt of Helios racks by year-end against a lo...
Reuters, via Tech Startups, reports capital released against deployment milestones with Anthropic deploying up to two gigawatts of Instinct MI450 starting 2027. Same structure as Nvidia/OpenAI: compute vendor capital flowing to the lab that commits to buy the silicon. A two-gi...
June local-inference benchmarks across the 128GB class put NVIDIA's DGX Spark (~$4k), AMD's Strix Halo / Ryzen AI Max+ 395 (~$2 to 3k), and the M5 Max 128GB (~$5k) head to head (Hardware Corner). Prompt processing favors CUDA hard. But token generation lands at a surprisingly...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.