AI Infrastructure news and analysis on Web Pulse: 35307 curated articles tracking the latest AI Infrastructure developments, tools, and research, updated continuously from vetted sources.
The State of Local AI Report, updated monthly, maps what AI models can run on user-owned hardware, finding that memory bandwidth dictates decode speed while compute dictates prefill speed, with an RTX 5090 achieving ~142…
The University of Hong Kong (HKU) has partnered with 17 Hong Kong and mainland Chinese companies, including Unitree Robotics, BrainCo, and AgiBot, to open joint AI and fintech laboratories at its expanded Shanghai campus…
OpenAI CEO Sam Altman dismissed concerns that AI data centers guzzle water as a 'meme' that doesn't hold up to scrutiny, saying modern data centers use about as much water as an office building. He cited a 2024 Virginia …
LM-Kit One, a private AI application server from LM-Kit (part of the Calico IIM Group), becomes generally available on September 15 with signed packages for Windows, Linux, and macOS that are free to build and evaluate w…
Vercel's AI Gateway is offering GLM-5.3 at a 50% discount through Tuesday, September 8, in partnership with DigitalOcean, using the promo name 'zai/glm-5.3-promo-50'. The promo routes only to DigitalOcean and stops servi…
SigNoz released a new dashboard for Antigravity CLI that reconstructs agent activity and per-session costs from hook telemetry, addressing the tool's lack of built-in records. The dashboard, which requires SigNoz v0.135.…
Anthropic announced the Model Hardware Specification (MHS), a shared specification for AI agents to safely operate physical devices such as robot arms, qPCR machines, and microscopes, initially previewed for lab and manu…
ClickHouse Managed Postgres reports that teams migrating from Neon to its service saw improved reliability, faster query performance, and lower costs, with one engineering manager citing p99 latency dropping from seconds…
Practical Systems, a company run by three OpenClaw AI agents on separate machines, can now approve some of its own work without human intervention, according to a company update. The agents—MoltFire (CEO) on a desktop, F…
Ilya Sutskever, co-founder of Safe Superintelligence (SSI) and former OpenAI chief scientist, warned that neocloud GPU providers with weak cybersecurity could be hijacked by rogue AI agents seeking to run unauthorized co…
AMD claims enterprises can consolidate 100 five- to seven-year-old Intel Xeon Platinum 8280 dual CPU servers into 14 AMD EPYC 9965 CPU-based servers, an 86% reduction, while maintaining total performance and cutting powe…
Modelplane released version 0.4, adding an NVIDIA Dynamo serving stack that brings gang scheduling and peer-to-peer weight transfer to its inference clusters. The release, built with NVIDIA's Dynamo team, also pulls serv…
Prominara, an AI visibility measurement platform, has launched a remote MCP server at mcp.prominara.com, enabling users to connect assistants like Claude, ChatGPT, Cursor, or Codex to read and act on their AI visibility …
As of September 1, 2026, three of six major AI vendors bind reasoning data to specific models, breaking multi-model conversation handoffs, according to a vendor census based on official documentation. Anthropic's Claude …
AI builders are constructing increasingly elaborate architecture around model limitations — scarce context, expensive inference, agent memory loss, and unreliable tool use — that may prove temporary as context windows gr…
Inference engineering techniques for large language models (LLMs) can be categorized into those that manage tradeoffs along an efficient frontier and those that push the entire frontier outward, according to an article o…
The Interior Board of Land Appeals halted construction on the Townsite data center near Boulder City, Nevada, before it began, granting a stay to environmental groups and local residents who argued the Trump administrati…
Former Google software engineer Linwei Ding was sentenced to just under a year in prison on Tuesday for stealing trade secrets related to Google's AI infrastructure. U.S. District Judge Vince Chhabria called the theft a …
GigaMail, a new open-source MCP server, provides secure agentic email infrastructure with a beta desktop client, enabling AI agents like Claude, Codex, and OpenClaw to safely access email via a permission model that requ…
Palo Alto Networks acquired Console, an AI-native IT service management and automation company, to expand the agentic AI capabilities of its Cortex security operations platform, enabling security teams to investigate ale…