AI Infrastructure news and analysis on Web Pulse: 34580 curated articles tracking the latest AI Infrastructure developments, tools, and research, updated continuously from vetted sources.
Twigg launched as a stateful LLM API that lets developers maintain chat context across multiple model providers without managing storage infrastructure themselves. The API works by returning a chat ID on creation, then a…
Apple is considering selling an AI server built on its "M8" series chips that would incorporate Nvidia networking hardware, The Information reported, potentially returning Apple to the server market it exited when it dis…
AI Model Radar published measurements on 16 September 2026 showing that the standard rule-of-thumb for estimating context-cache memory in open-weights models can be off by a factor of four, and by roughly forty times for…
Apple is developing enterprise servers powered by M8 Ultra chips and plans to sell the AI-focused machines rather than use them only internally, according to a new report. Two versions are said to be in development, with…
Maxim AI has released Bifrost, an open-source AI gateway written in Go that acts as a runtime control plane for enforcing AI risk assessments and governance policies. The tool is positioned to help enterprises map contro…
Nvidia, Google, and Emerald AI announced the formation of the AI Energy Management Alliance, a collaborative initiative to optimize electricity use in AI data centers by letting facilities adjust power consumption based …
A company is soliciting feedback on a planned local AI workstation build featuring 1-2 Nvidia RTX Pro 6000 GPUs, an AMD Threadripper Pro 9955WX CPU, 64GB of DDR5 memory, an Asus Pro WS WRX90E-SAGE SE motherboard, a Seaso…
Speechify released Simba 3.2 and Simba 3.0, text-to-speech models that cut time-to-first-audio to roughly 100ms on Coval's benchmark while keeping the same voices, quality, and per-character pricing. The company said the…
A white paper presented by IEEE Spectrum and Wiley and sponsored by CoolIT Systems argues that single-phase direct liquid cooling can handle AI and high-performance computing thermal loads for the next decade, as modern …
Huawei unveiled a four-layer "3D data center" in Wuhu, Anhui province, on Tuesday, an architecture the company says is designed for AI supernode clusters containing more than 100,000 accelerator cards. Huawei Deputy Chai…
Bifrost, an open-source AI gateway written in Go by Maxim AI, provides a centralized control plane to govern Model Context Protocol (MCP) tool execution and block known vulnerabilities. The project details how direct poi…
Meta is promoting the cost-saving and energy-efficiency benefits of its custom in-house AI processors as it scales infrastructure for advanced machine learning models, with industry analysts estimating custom chips can c…
A user running Qwen3.8-Flash-Next at IQ3 quantization on a system with 64GB of RAM, an AMD 9950X3D, a 9070XT, and a ZFS stripe of three mid-range NVMe drives reported fitting 255K tokens of context using buun-llama's VBR…
Construction has begun on a data hub at the site of the Shippingport nuclear power plant in Beaver County, Pennsylvania, the first commercial nuclear power plant in the United States, which opened in 1957. The Washington…
Apple is developing enterprise servers, according to a report from The Information, expanding its custom server hardware program beyond internal AI workloads and into a market dominated by Dell and HPE. Apple has been sh…
Cornelis Networks announced Active Compute Fabric, a networking architecture that moves programmable compute into the switch, alongside a $205 million funding round on September 14. The company's CN5000 400 Gbps switch b…
Uber cut cost per 1,000 AI requests by roughly 34% and cost per session by about 52% from peak while weekly active users of its AI coding tools rose 7x and weekly agent requests climbed 9.4x between February and August 2…
A benchmark of llama.cpp v0.4.1-dev (commit 83078fe) on an Intel Panther Lake laptop found the OpenVINO backend unable to run the Qwen3.6-35B-A3B-MTP-UD-Q4_K_M.gguf model, failing with a quantized-V-cache flash-attn erro…
Lenovo announced the ThinkAgile VX850 V4, Premier Deploy for Infrastructure, and expanded Express Solutions to help enterprises modernize virtualized environments for AI workloads. The ThinkAgile VX850 V4 targets highly …
Emerald AI, Google and NVIDIA announced the launch of the AI Energy Management Alliance (AEMA), a coalition aimed at advancing data centers that can dynamically manage electricity use in response to grid conditions. AEMA…