AI Chips news and analysis on Web Pulse: 6895 curated articles tracking the latest AI Chips developments, tools, and research, updated continuously from vetted sources.
Nvidia announced at the OCP Global Summit in San Jose that future BlueField data processing units and ConnectX-10 SuperNICs will support Google's Falcon hardware transport protocol, with Falcon now supported on Nvidia Sp…
Nvidia confirmed its Vera CPU offerings have begun shipping in "multiple form factors," with dedicated Vera racks "shipping now," according to Joe DeLaere, Nvidia's product lead for AI infrastructure solutions. DeLaere s…
NVIDIA contributed its Vera Rubin rack and compute tray designs to the Open Compute Project through the open NVIDIA MGX architecture at the OCP Global Summit in San Jose, giving system builders a common foundation for ne…
NVIDIA and Cisco announced plans for Ultra-Dense Small Form Factor Pluggable (USFP), a new pluggable optical module designed for 400Gb/s-per-lane electrical signals that enables AI factory builders to reach up to 12.8Tb/…
Microsoft unveiled the Maia 300 AI accelerator on October 12, 2026, its third-generation in-house inference chip, and said it will scale the Maia family to "multi-gigawatt deployments" from 2027 as "the primary inference…
Marvell Technology unveiled its second-generation Structera X CXL memory expanders, the 3406 and 3506, on October 12, 2026, claiming they are the industry's first devices with six memory channels for CXL 3.2 — a 50% incr…
AMD unveiled Helion, a CPU-only rack-scale system for agentic AI workloads, at the OCP Global Summit in San Jose on October 12, with CEO Dr. Lisa Su announcing the design during her keynote. Helion is powered by 6th Gen …
New cooling technologies spreading through AI data centers reduce or eliminate on-site water use but can raise total water consumption by increasing electricity demand 6-35% during peak conditions, according to Bloomberg…
A Ukrainian drone strike on October 11 knocked Yandex's 50-megawatt Vladimir data center completely offline, the third of the company's publicly known supercomputer-class facilities disrupted in four days after strikes o…
A research report finds that robot post-training consumes roughly three to four times the GPU resources of text-only post-training at equal parameter count, driven by video, joint and action signals that carry far more i…
AMD introduced the AMD Helion agentic AI rack on Oct. 12, 2026, an Open Compute Project spec-design blueprint that supports CPU-only configurations of up to 32,768 cores on 6th Gen AMD EPYC processors and all-in-one rack…
Microsoft announced at a Windows and Surface event in San Francisco on October 7 that Copilot on Copilot Plus PCs can now read local files and recent activity with permission, take actions across Windows such as organizi…
Microsoft unveiled Windows hybrid intelligence at a Windows and Surface event in San Francisco on October 7, an architecture that routes each AI task between on-device chips and cloud data centers based on cost, speed an…
Five of six Qwen3.8-family model runs on AMD RDNA4 GPUs with the Radiance inference engine produced a playable Breakout game from a single 6,567-token prompt with no harness or retry loop, according to a Level1Techs writ…
A developer published a guide showing how to cap AI agent GPU spending at the platform level rather than in prompts, citing an example where a forgotten 8-GPU H100 run from Friday evening to Monday morning would cost $1,…
A Level 1 Techs forum thread weighing used Nvidia V100 GPUs against AMD Strix Halo machines for local AI found V100-16GB cards selling for $800-$1,000 each, making a two-card build cost at least $3,000, while Minisforum'…
Microsoft announced the Surface RTX Spark Dev Box at its October 7 Windows and Surface event, a $5,999 fanless compact desktop that runs models exceeding 120 billion parameters locally using 128 GB of unified memory on N…
Apple has confirmed that its iOS 18.4 beta embeds generative AI directly into the device's neural engine, running on the A18 and M4 chips' NPUs rather than in the cloud. The update introduces a "Private Compute" framewor…
Robotic systems are now assembling NVIDIA Grace Blackwell GB300 tester trays, shifting high-precision physical labor from human technicians to machines that combine robot learning with tactile feedback, according to a re…
President Trump signed Executive Order 14420 on August 26, declaring a national emergency over foreign-made equipment in the U.S. bulk-power system and giving the Department of Energy authority to block transactions invo…