AI Chips news and analysis on Web Pulse: 6915 curated articles tracking the latest AI Chips developments, tools, and research, updated continuously from vetted sources.
OpenGEMM, an open-source project from developer aramesh10, released CUDA GEMM kernels for NVIDIA's B200 GPU (sm_100a), installable via pip and requiring PyTorch 2.8+ and CUDA 12.9+. The library emits standalone CUDA kern…
AMD and Intel's joint AI Compute Extensions specification, ACEv1, was merged into the GCC 17 development branch on September 2, 2026, delivering up to 16 times the compute density of equivalent AVX10 operations. The spec…
Americans for Responsible Innovation is urging House and Senate Armed Services leaders to overhaul AI chip export oversight in the FY2027 National Defense Authorization Act, citing continued Russian military acquisition …
Senator Chris Coons urged President Trump to prioritize artificial intelligence cooperation with Chinese President Xi Jinping, comparing the need for bilateral AI safety talks to Cold War nuclear arms negotiations during…
Thieves hijacked AI hardware shipments in California, according to a report whose body contained only subscription and membership promotional text with no further details. No specific figures, dates, companies, or named …
Anthropic CEO Dario Amodei published an essay, "We Must Pace the Frontier," calling for a measured slowdown in frontier AI development paired with measures to preserve the U.S. lead over China, and Anthropic announced it…
A reproducible Windows CUDA compatibility stack built on ZLUDA v6-preview.69 and AMD HIP SDK 6.4 has been published on GitHub, validated only on the AMD Radeon RX 9060 XT (gfx1200). The setup passes cuda_check for nvcuda…
Alex Ziskind built a four-GPU workstation with 384GB of VRAM using four RTX Pro 6000 Blackwell cards and a 32-core Threadripper Pro 9975WX, and measured that 71% of every coding-agent turn never touches the GPUs, conclud…
Researchers at Xi'an-based Xidian University, working with City University of Hong Kong and Fudan University, demonstrated more than 10 billion writing cycles in wurtzite ferroelectrics, roughly 100 times the endurance p…
The FBI, the National Security Agency, and the Cybersecurity and Infrastructure Security Agency alleged Tuesday that DeepSeek, Moonshot, and four other Chinese AI companies extracted "capabilities worth billions" since 2…
Empyrean Technology, China's top electronic design automation (EDA) tool maker, cut a circuit layout task from four weeks to one week by applying an AI agent to its simulation and layout tools, chairman Liu Weiping said …
OpenAI paused new $200/month Pro tier subscriptions for its GPT-6 Astra model within 48 hours of previewing it, as test-time compute and inference scaling from autonomous reasoning agents overwhelmed GPU cluster capacity…
South Korea's Ministry of Trade, Industry and Resources plans to build AI data centers and related infrastructure in the Honam region, with SK committing 70 trillion won to a 1-gigawatt AI data center and Samsung SDS pla…
South Korea's Jeonnam-Gwangju Integrated Special Metropolitan City aims to begin first-phase construction of a southwest semiconductor production base next year, complete it by the end of 2029 and start initial semicondu…
Dennis Bloch, counselor for economic affairs at the German Embassy in Seoul, told the Jeonnam-Gwangju Integrated Special Metropolitan City Semiconductor and AI Transformation Global Forum in Gwangju on Thursday that the …
Hyundai Motor Group's autonomous driving subsidiary 42dot will debut an Nvidia-powered Level 2+ vehicle fleet in 2028 and launch its own Atria AI-powered Level 2++ fleet in the second half of 2029, CEO Park Min-woo said …
Apple unveiled its third generation of Apple Foundation Models (AFM), a family of five models built in collaboration with Google, spanning two on-device models and three server-based models running on Private Cloud Compu…
D-Matrix announced it will bring its next-generation XPUs to NVIDIA's NVLink Fusion platform, scaling the company's Raptor accelerators beyond a single XPU into MGX racks. The deployment also uses NVIDIA's Vera CPUs, Con…
Cerebras is now serving OpenAI's GPT-5.6 Sol at up to 750 output tokens per second, according to a Cerebras blog post, a speed the author says shifts the bottleneck from the model to the surrounding agent harness. The pi…
High bandwidth memory (HBM) is hitting scaling limits driven primarily by thermal gradients across the stack, according to a technical deep dive that cites JEDEC standards and Samsung's HBM4 architecture presented at ISS…