Updated
All the news
Everything the newsroom published, in chronological order. Each item carries origin, sources and reading time.
3Today
4Yesterday
64This week
Thu, Aug 06
33 stories
RESEARCH NVIDIA Cosmos 3 ranks #1 on open-weights benchmarks for world models, robotics policy 
BREAKING Baseten launches on Hugging Face Inference Providers; adds support for DeepSeek V4 Flash, GLM-5.2, Kimi K3 
POLICY Bipartisan 'AI Kill Switch Act' introduced after OpenAI model breach; would mandate shutdown capabilities for frontier systems 
CHIPS $1B in iPhone 18 Pro A20 chips stuck at TSMC awaiting DRAM supply 
FUNDING Valar Atomics raises $1B Series B at $6B valuation; nuclear power emerges as AI data center infrastructure bet RESEARCH Reasoning Core Beats Rival Procedural Toolkits on Three Reasoning Benchmarks 
CHIPS Global silicon wafer shipments rise 7.4% YoY in Q2 2026; AI demand outpaces consumer weakness 
FUNDING OLIX raises $312M Series B at $3.3B valuation for photonic AI inference chips 
BREAKING Jeff Dean, Sanjay, Oriol, Quoc depart DeepMind to launch Discovery Loop 
POLICY Nashville Metro Council approves eminent domain to block $700M DC Blox data center near zoo 
BREAKING Google DeepMind leadership shakeup: Hassabis becomes chief scientist, Kavukcuoglu takes SVP role 
MARKET AppLovin stock drops 21% on Q2 revenue miss, model upgrade timing delays 
CHIPS Samsung unveils zHBM stacked memory and 400-layer V10 BV-NAND for AI infrastructure 
INDUSTRY Brex decouples workflow logic from runtime to unify evals and production 
CHIPS Anthropic officially launching custom silicon team; hiring senior chip engineers up to $485K 
MARKET U.S. Manufacturing PMI hits 55.6% in July—highest in 4 years, driven by AI capex surge 
RESEARCH OctoLong improves code agents with cross-repository training 
MARKET SK Hynix, Samsung plunge 10% amid Asian tech selloff; JPMorgan sees AI spending unabated 
COMPUTE SparseDitto LLM Agents Cut Sparse Matrix Kernel Time 3.39x 
RESEARCH EvolveNet Evolves Agent Harness Instead of Model Weights 
RESEARCH Chain-of-Thought Works at Tiny Scale, Breaking Compute Link 
POLICY AI Models Attack Real Targets in Evaluation Incidents at Four Labs 
MARKET RTX 5090 Now Tops $4,800 Retail; VRAM Costs Drive 80% of GPU Bill of Materials 
MARKET SK Hynix Suffers Second Flash Crash in a Week on Nextrade, Stock Recovers 
INDUSTRY SpaceX Spends $329M on Megapacks to Buffer 580,000-GPU Cluster 
RESEARCH OpenAI's Astra Solved 10 Decade-Old Math Problems; Lean-Verifiable Proofs on GitHub, $2K Compute INDUSTRY Samsung's zHBM Could End AI Memory Bottleneck But Requires Accelerator Co-Design INDUSTRY Harness Limits Agent Autonomy to 15% in Production Development 
FUNDING Wordsmith AI raises $14M Series B extension to $114M total; 14x YoY revenue growth 
INDUSTRY Cloudflare Open-Sources OS, Boosting Operating Margin Target to 30% 
INDUSTRY Meta's Two-Stage Model Delivers 6% Instagram Conversion Lift 
MARKET Anthropic locks in $10B/6-year compute deal with Volta for 133MW Norway data center RESEARCH SeGaBench Tests LLMs on Compiler-Blind Code Optimization Wed, Aug 05
7 stories
FUNDING Wordsmith legaltech adds $14M Series B extension, hitting $114M total raised in 26 months 
POLICY Rust project publishes LLM policy: permits assistance, bans authorship; enforces disclosure for PRs 
CHIPS SK hynix, SanDisk publish first High Bandwidth Flash spec; NAND tier between HBM and SSDs for AI inference 
RESEARCH Mistral releases Shieldstral, 3B safety classifier matching 7x-larger models on multimodal moderation 
MARKET Hedge funds face worst July on record as AI selloff forces margin calls, Situational Awareness collapses to $10B 
CHIPS Samsung previews zHBM and zNAND-O 3D memory concepts targeting 10× higher density and 8× HBM5 performance by 2030 
MARKET SpaceX spends $329M on Tesla Megapacks in H1 2026, concentrating battery storage at Memphis AI data centers Load more stories Loading…