AI Infrastructure news and analysis on Web Pulse: 35824 curated articles tracking the latest AI Infrastructure developments, tools, and research, updated continuously from vetted sources.
YMTC, China's largest NAND flash manufacturer and the third-largest global supplier by shipments, is preparing an IPO on the Shanghai STAR market to fund its pivot to high-value AI enterprise SSDs. The company has begun …
Anyscale and NVIDIA cuDF have integrated GPU-native operators into Ray Data, enabling up to 3x better total cost of ownership (TCO) on select data curation workloads compared to CPU-based solutions. The new features, ava…
Anyscale's Ray team reports that after a round of Ray Core improvements, batch inference on 500 nodes is 23% faster end to end, Ray Data shuffle is 24% faster, and placement groups are 62× faster to ready at 2,000 nodes …
Ray Serve LLM, developed by Anyscale, introduces token-load-aware routing to improve LLM serving efficiency, moving beyond naive KV cache reuse strategies. The new approach addresses the challenges of stateful execution,…
Anyscale co-founder Christian Stano outlines a three-step maturity curve for companies to build 'learning loops'—compounding flywheels of data curation, custom model training, and inference powered by proprietary data—as…
SkyRL now supports FP8-accelerated training and rollout with on-policy weight sync, reducing end-to-end step time by up to 23% while closely tracking BF16 convergence in long-run RL experiments. The system, developed by …
Ray 2.58 introduces native sandboxing built on gVisor, allowing users to create isolated environments from OCI container images and manage them with Ray primitives. The feature, developed in partnership with Google, scal…
Nebius, a Nasdaq-listed AI cloud platform headquartered in Amsterdam, is hiring a Data Engineer for its agentic search product Tavily, based in Israel. The role involves building and scaling a data platform that ingests …
Taiwanese prosecutors indicted nine people, including employees of Nvidia and Super Micro Computer, for illegally exporting AI servers to China, with 50 servers transshipped through Indonesia, eight via Japan, and the re…
Nvidia's LPX racks for AI inference accelerators have entered full production, housing 256 Groq 3 language processing units (LPUs) interconnected with 640 terabits per second (Tb/s) of scale-up bandwidth. The platform, w…
Ablo, a new infrastructure platform, aims to make AI applications multiplayer by enabling multiple people and agents to read and write on the same state, with governance over who can act on it. The platform addresses the…
Nvidia is reportedly paying $6 billion for a non-exclusive license to Poolside's Model Factory, offering jobs to 109 employees, and investing an additional $1 billion in Poolside at a $12 billion pre-money valuation, acc…
An engineer has built Timeline Studio, an open-source, local-first AI video editor that runs entirely in the browser, using WebGPU, WASM, and ONNX to process media and AI models locally instead of uploading to a server. …
A developer built react-native-needle, a React Native binding for the Needle 2 on-device LLM, and found that while it excels at tool-calling tasks with sub-300ms response times, it can fabricate numbers with 100% confide…
A developer exploring the execution-state-preflight reference skeleton on GitHub concludes that the model should not own the authority to decide that an execution input is settled, proposing that 'unknown' be produced by…
A developer's portfolio agent answers questions using only 0.598% of its 1,003 indexed chunks, yet reports answers with the same confidence as if it had read the entire corpus, a flaw the developer says is common in retr…
SK hynix is collaborating with TSMC and Intel Foundry to integrate its HBM4 memory into advanced packaging technologies, including Intel's EMIB, as disclosed at Hot Chips 2026. The collaboration aims to support next-gene…
ClaudeGate, a new open-source local API gateway, enables Claude Code CLI and Anthropic SDK applications to use any AI model, including free and frontier models from OpenRouter, DeepSeek, and others, by translating Anthro…
OpenAI has secured a $122 billion funding round, positioning it among the highest-valued private AI companies, as CEO Sam Altman described AI compute as 'the most expensive infrastructure project in history.' The company…
Alibaba launched Wan3.0 on August 24, a video generation API that accepts documents (PDF, DOC, XLS, PPT, TXT, MD, and webpage URLs) and up to 20 multimodal assets per request, producing 30-second 1080p videos with native…