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.
NVIDIA Research's Efficient AI Team and Singapore Lab reported that a two-stage pipeline specialized for a single NVIDIA DGX Spark generates a 384p draft and refines it to 768p, producing a 5-second 768p video in 56 seco…
Researchers submitted a paper to arXiv on 24 Feb 2026 introducing OpenGCRAM, a memory compiler that supports both SRAM and Gain Cell RAM (GCRAM) and generates macro-level designs and layouts for commercial CMOS processes…
Apple and TSMC unveiled the A20 Pro on September 9 at Apple's "Surprise and Shine" event, the first iPhone chip built on TSMC's 2 nanometer (N2) process node, which will power the iPhone 18 Pro, iPhone 18 Pro Max, and th…
The US Department of Justice has launched an antitrust probe into Nvidia's $20 billion acquihire of Groq, The New York Times reported this week, but The Register reports that unwinding the deal would likely change little…
Developer Reindert Pelsma released nvkvm-pv v0.2.5, an experimental open-source project that lets KVM virtual machines run unmodified CUDA, PyTorch and Vulkan workloads on an NVIDIA GPU while the Linux host keeps using t…
Intel released Intel Linux NPU Driver 1.38 on Friday, adding official support for Ubuntu 26.04 LTS as the sole listed change in the user-space driver for Core Ultra Neural Processing Units. The driver stack already worke…
TSMC plans to double its CoWoS advanced packaging capacity to 260,000 wafers per month by 2028, up from an estimated 130,000 by the end of 2026, according to 科技新報 TechNews. The current CoWoS bottleneck is pushing overflo…
The Register's daily news roundup highlights Nvidia's Groq acquihire drawing DOJ scrutiny, with the outlet noting that even unwinding the $20B deal would leave a growing list of alternatives ready to take Groq's place wi…
A forum poster clarified that their original proposal was not to use an NVMe SSD as VRAM, but to design future consumer GPUs with an expandable, lower-cost memory tier managed directly by the GPU, similar in concept to H…
Tesla has built the largest grid-connected battery system in the United States at xAI's Colossus 2 data center in Memphis, Tennessee, using 720 Megapacks with an estimated storage capacity of roughly 2.8 GWh, according t…
DeepSeek released V4.1 Flash, a 552-billion-parameter mixture-of-experts model with 8 billion active input and 16 billion active output parameters that scored 74.2 on Deep SWE and 30 on Terminal Bench 3.0, placing it nea…
Sustained AI inference workloads on Nvidia RTX 3090 and 4090 cards accelerate thermal paste pump-out, a failure mode in which paste migrates off the GPU die and causes clock throttling even while reported die temperature…
A new proposal calls for a common GPU execution and memory-management layer above CUDA, ROCm, and Intel XPU/oneAPI so that AI workloads are not tightly coupled to a single vendor, targeting 13B-class INT4 inference on 12…
OpenAI is hiring a Research Engineer for AI for Chip Design in San Francisco at a salary of $380,000 to $500,000 per year, a range the job board reports sits 93% above the $228,000 median for AI Agents roles in the Unite…
OpenAI has paused new subscriptions and upgrades to its $200-a-month ChatGPT Pro 20X plan after demand for its Astra model strained computing capacity, OpenAI product executive Thibault Sottiaux said. Existing Pro 20X cu…
A Proof-of-Concept repository published on GitHub under the project jax-softmax-bypass claims to bypass the Transformer Softmax's transcendental exponential function through a static algebraic contraction framework that …
Dave Ditzel, RISC pioneer and founder and former CEO and CTO of Esperanto Technologies, delivered a keynote and AMA on August 25, 2026, in Palo Alto at the Cool Chips - Hot Takes gathering organized by AiNEKKO, AIFoundry…
Consumer-facing agentic AI products such as xAI's Grok Bot, Meta's Muse, and Instinct could push adoption from early adopters to the early majority and drive demand for server CPUs, according to an analysis of the AI inf…
NVIDIA announced the Jetson Orin Nano 2 on August 25, a robotics computer delivering 78 TOPS and roughly 2x inference throughput over the Orin Nano Super at 40% lower power, with the module and developer kit shipping in …
Rapidus Corp President and CEO Atsuyoshi Koike said Thursday he is seriously considering building a semiconductor plant on the moon around 2040, citing lunar water resources needed for chip production. "This is not a jok…