AMD launched the Instinct MI400 Series data-center GPU family, Helios rack-scale solution, and CDNA 5 architecture at its Advancing AI 2026 keynote, with the flagship Instinct MI455X featuring 320 bil…
AMD has introduced the EPYC 9006 'Venice-LP' processor, a purpose-built enterprise CPU for rack-scale AI solutions that competes directly with NVIDIA's Vera CPU. The chip features a 72-core/144-thread…
AMD at its Advancing AI 2026 event in San Francisco detailed its next-generation AI infrastructure, including the Instinct MI455X GPU, 6th Gen EPYC "Venice" CPUs, and the Helios rack-scale platform. T…
AMD introduced the Instinct MI455X GPU with 432GB of HBM4, a 50% increase over NVIDIA's B300 or Rubin, and the 72-GPU Helios rack achieving 2.9 exaFLOPS of FP4, at its Advancing AI event. The company …
The llm-d project introduces co-operative time-slicing to interleave independent reinforcement learning jobs on shared hardware, increasing aggregate accelerator duty cycles from ~40% to 70% without i…
Triton-blackhole, a deterministic numerical debugger for Triton kernels, identifies where output diverges from a PyTorch reference and classifies drift as benign fp16/bf16 drift or a real bug without …
AMD's Advancing AI 2026 event highlighted a strategy to bypass NVIDIA's CUDA moat through architecture, as argued by Chris Lattner, creator of LLVM, who also claimed AI is "mid" to emphasize that the …
NVIDIA and Prime Intellect Lab have partnered to enable developers to customize the NVIDIA Nemotron 3 Nano 30B-A3B model using reinforcement learning with verifiable rewards on the Python Math environ…
White House Office of Science and Technology Policy Director Michael Kratsios accused Chinese AI startup Moonshot AI of illegally accessing NVIDIA GB300 chips in Thailand and conducting covert distill…
CamelAI self-hosts DeepSeek V4 Flash on AWS spot instances to power its free tier, using g7e.24xlarge instances with four RTX PRO 6000 GPUs. The setup keeps free-tier costs fixed by avoiding per-token…
Graph engineering, the latest buzzword for orchestrating multiple agent loops, forces teams to formalize processes but was already covered by Anthropic in 2024, according to Louis-François Bouchard, C…
AMD CEO Dr. Lisa Su will deliver the Advancing AI 2026 keynote in San Francisco, unveiling the company's first rackscale system, Helios, which integrates MI455X GPUs, EPYC 'Venice' CPUs, and Pensando …
Framework previewed a 192GB configuration of its Desktop powered by AMD's Ryzen AI Max+ Pro 495, offering 50% more memory than the previous 128GB model but only 7% more bandwidth, at AMD's Advancing A…
Nunchaku integrates 4-bit diffusion inference into the Diffusers library, enabling large models like Flux.1 to run on consumer GPUs with significantly reduced VRAM requirements while maintaining image…
AWS raised H200 GPU instance prices by 15% on January 4, 2026, the first GPU price increase in roughly two decades, while average GPU utilization across more than 23,000 Kubernetes clusters sits at ju…
NVIDIA CEO Jensen Huang highlighted the resilience of Chinese AI, noting that despite hardware constraints, the region's developers are achieving rapid iteration in prompt engineering and model optimi…
IREN Limited on July 20 disclosed $2.8 billion in new multi-year AI cloud contracts with leading AI developers, raising its year-end 2026 annualized run-rate revenue target from $3.7 billion to over $…
Average GPU utilization in Kubernetes clusters sits at just 5%, according to the Cast AI 2026 State of Kubernetes Optimization Report, with AKS at 2%, EKS at 5%, and GKE at 6%. An idle H100 GPU on AWS…
NVIDIA released Cosmos 3 Edge at SIGGRAPH 2026, a 4-billion-parameter open world model for physical AI that runs on a consumer GeForce RTX GPU, with weights available on Hugging Face and commercial de…
Poolside AI co-founder Eiso Kant says the company has built a Model Factory that can take a model from pre-training to release in eight weeks, running 10,000–20,000 experiments per month with fewer th…