AI Infrastructure news and analysis on Web Pulse: 35428 curated articles tracking the latest AI Infrastructure developments, tools, and research, updated continuously from vetted sources.
A technical explainer from an unnamed author details four distinct caching layers in LLM inference—KV cache, prefix caching, prompt caching, and semantic caching—clarifying that only the semantic cache uses fuzzy matchin…
Snowflake AI Research introduced Semi-Persistence, a model-serving approach that keeps model weights in persistent CPU memory to enable fast swapping of GPU copies, reducing end-to-end sleep and wake-up cycles by 5.6× to…
Meryem Arik, co-founder of inference company Doubleword, said most companies overpay for AI inference by 2x to 5x, sometimes by an order of magnitude, because their inference stacks do not match their use case priorities…
A technical analysis by an unnamed author argues that nondeterminism in LLM inference is not primarily caused by GPU concurrency and floating-point non-associativity, as commonly hypothesized, but by other factors such a…
Adnan Masood, writing for InfoWorld, argues that AI governance should be treated as engineering work, encoding policies into executable controls that enforce access, constrain actions, capture evidence, and measure drift…
TSMC's Dr. Ming-Fa Chen presented updates on the COUPE silicon photonics platform at SEMICON Taiwan 2026, detailing plans to integrate modulators, light sources, and Semiconductor Optical Amplifiers (SOA) onto the platfo…
Solid-state transformers (SSTs) are emerging as a key technology for next-generation AI data centers, converting medium-voltage AC directly to 800 V DC to improve efficiency and reduce equipment footprint. Delta, a maker…
NVIDIA has released the NVIDIA Inference Xfer Library (NIXL), an open-source library designed to accelerate point-to-point communications in AI inference frameworks such as NVIDIA Dynamo, with support for Linux environme…
Japan's Defense Ministry on Monday disclosed an 8.9 trillion yen ($55.6 billion) budget request for the next fiscal year, which includes plans to deploy attack drones, develop hypersonic missiles, and integrate an AI pla…
OpenAI and Anthropic have purchased tens of thousands of Apple Mac minis and Mac Studios to train computer-use agents, according to The Information. The demand has been so high that the most powerful models have been sol…
Perplexity is preparing Hybrid Mode for Perplexity Computer on Mac, allowing the agent to orchestrate in the cloud while delegating lighter subtasks to a local model, with three downloadable local models including a reco…
Sembcorp, a leading energy and urban solutions provider, said the global push to lead in artificial intelligence is becoming a race to secure energy, with the International Energy Agency projecting AI facilities' electri…
Singapore will invest S$220 million over three years in the fourth tranche of the Financial Sector Technology and Innovation (FSTI) scheme, announced by the Monetary Authority of Singapore (MAS) on Aug 31. The funds will…
Google announced Gemini 3.7 Flash just three weeks after its previous release, according to Ars Technica. OpenAI shared early results for its Jalapeño inference chip, claiming better performance per watt and lower latenc…
OpenAI unveiled Jalapeno, its first custom inference chip co-developed with Broadcom, at the Hot Chips conference, reporting independent benchmarks showing it outperforms Nvidia's GB200 and GB300 systems by up to 1.9x in…
HAProxy's website is protected by Anubis, a proof-of-work challenge system designed to block AI companies from scraping the site, which the administrator says causes downtime and makes resources inaccessible. The system,…
SZG Labs, an enterprise engineering firm, argues that the era of chatbot-based AI is ending and that businesses should shift to agentic workflows, where AI systems execute tasks autonomously rather than merely answering …
OpenAI's ChatGPT Work Cloud now offers unrestricted internet access in its code-execution sandbox, enabling direct API calls, package installs, and configurable domain access, unlike regular Chat or Claude. The feature i…
Advanced cooling technologies are needed as modern vehicles with electric drivetrains, ADAS, and AI processors generate heat fluxes that challenge traditional cooling architectures, according to an article from EE Times.…
Kosha, an open-source storage-disaggregated search engine built in Rust as an alternative to Elasticsearch/OpenSearch, has completed Phase 1 with BM25 lexical and kNN/ANN vector search implemented across all seven crates…