AI Chips news and analysis on Web Pulse: 6900 curated articles tracking the latest AI Chips developments, tools, and research, updated continuously from vetted sources.
A developer evaluated speculative decoding in vLLM on AMD Instinct MI300X and MI355X GPUs, testing draft methods including MTP, EAGLE-3, DFlash, and DSpark that propose candidate tokens verified in a single forward pass.…
Wendell built a custom AI workstation PC for a client who wanted to stop paying for expensive tokens, using a SilverStone Alta T1 case, an Asus Pro WS W890E-Sage SE motherboard, an Intel Xeon 658X CPU, and RTX 6000 GPUs.…
Federal agents arrested Greg Lui, 38, also known as "Yiu Kong Lui," of San Gabriel, California, on Wednesday on charges that he used his company Earthmade Computers to smuggle more than $300 million in restricted Nvidia …
NetApp announced NetApp Novus at its INSIGHT 2026 conference in Las Vegas, a disaggregated storage architecture for AI factories that separates metadata from the data path and that NetApp says is designed to exceed 100 T…
A user on the Level1Techs forum reported getting four AMD R9700 AI GPUs working behind a PLX88096 PCIe switch on an AM4 X570-D4U-2L2T/BCM motherboard running Proxmox, after enabling Legacy CSM support resolved the initia…
CoreWeave announced at its Fully Connected conference in San Francisco that Nvidia's Vera Rubin NVL72 rack-scale AI platform is available in its cloud, with Cognition, developer of the Devin software-engineering agent, a…
A firm rented four Nvidia H200 GPUs to test DeepSeek's claim that its model runs 80 times cheaper, according to the article's headline. The source body contains no further reporting, figures, or named organizations beyon…
Qualcomm announced an early developer preview of Linux running on its upcoming Snapdragon X2 Series laptop processors, revealed alongside demonstrations at Snapdragon Summit, targeting kernel contributors, toolchain deve…
Synopsys launched GPT-Synopsys, a specialist chip-design AI model built with OpenAI, during its Investor Day on Wednesday as part of a revenue-sharing agreement. Synopsys CEO Sassine Ghazi said the model will help more c…
NVIDIA says its AI factories are engineered to maximize return on investment across three dimensions — earning capacity, useful life and demand — citing SemiAnalysis AgentX data that NVIDIA Vera Rubin NVL72 systems deliv…
Google announced Gemini 4 Argon on September 30, a model that leads or ties GPT-6 Astra and Claude Opus 5.5 on 14 of 19 Google benchmarks and scores 53 on Artificial Analysis, level with Astra but below Opus 5.5 at 58 an…
A technical explainer details how KV cache memory, not model weights, dominates VRAM during LLM inference: for a 13-billion-parameter model with 40 layers, 40 KV heads and 128-dimension heads at 16-bit precision, the KV …
A new attributed record of AI compute infrastructure catalogs 83 facilities, 482 GPU clusters and 2,935 price observations, finding that official H100 rental prices range from $4.57 to $18.37 per accelerator-hour across …
OpenAI and chip design software maker Synopsys announced a multi-year partnership on Wednesday to build GPT-Synopsys, a specialized model trained to operate Synopsys's electronic design automation tools as what the compa…
NVIDIA announced the general availability of its cuObject client and server libraries and expanded the xio-sig project to include cuObject alongside cuFile, in partnership with Google Cloud and Microsoft. The company als…
OpenAI's hardware team, led by Richard Ho, has disclosed measured performance for Jalapeño, its first custom inference accelerator built with Broadcom as design partner and Celestica handling board and rack integration. …
Chutes AI trained an 8 billion parameter model on 240 RTX 5090 consumer GPUs spread across 30 hosts in 13 countries for approximately $6,500, or about $11 per billion tokens processed, Jon Durbin said in a presentation o…
A user on an AM4 X570-D4U-2L2T/BCM motherboard resolved missing PCIe BAR space that blocked four R9700 AI GPUs behind a PLX88096 switch by disabling Legacy CSM support in the BIOS, which also restored the onboard 10GbE c…
NetApp unveiled Novus, a disaggregated storage architecture it says can exceed 100 TB per second of aggregate throughput and support environments with hundreds of thousands of GPUs, though the performance claim has not b…
Chinese AI labs' KV cache optimizations, led by DeepSeek's MLA architecture and its DeepSeek-V4.1-Flash release, cut the global KV cache to 890 bytes per token and reduced the cache footprint for long-context use cases s…