AI Infrastructure news and analysis on Web Pulse: 35802 curated articles tracking the latest AI Infrastructure developments, tools, and research, updated continuously from vetted sources.
Researchers from UT Austin and UC Berkeley released FreeToken, an edge-native Mixture-of-Experts (MoE) serving engine that adapts model execution to available hardware bandwidth, enabling a 753B GLM-5.2 model to run on a…
At NVIDIA's GTC conference, Adobe CTO Ely Greenfield revealed that early AI training runs wasted roughly two-thirds of GPU time waiting for data, costing about $600,000 of every $1 million spent on training. Adobe fixed …
A developer built CarWatch, a Raspberry Pi 5-based in-car AI assistant that runs a 35B-parameter Qwen3.6-35B-A3B model locally at 3.5 tokens per second, achieving full offline functionality for voice, owner's manual quer…
Perplexity AI launched Portable Computer, a local-first AI agent that runs on NVIDIA's DGX Spark hardware and processes tasks on-device before reaching the cloud, available now for Pro and Max subscribers. The feature ha…
Emerald AI, a startup that helps AI data centers reduce power use during grid strain, raised $150 million in Series A funding, valuing the company at $1.05 billion. The round, one of the largest for an AI infrastructure …
OpenAI said its Broadcom-built Jalapeño inference chip delivers 1.5 to 1.9 times more AI work per watt than Nvidia's GB200 and GB300 systems on tested models, with 1.7 to 3.6 times lower end-to-end latency, based on benc…
Apple refreshed the Mac mini and Mac Studio with four chips, including the M6, its first 2nm chip, and the M5 Ultra, the first quad-die Apple silicon, both aimed at running AI models locally. Pre-orders open today, with …
A developer argues that multi-agent AI systems should be built as event-driven architectures using message brokers like RabbitMQ or Kafka rather than synchronous HTTP chains. The approach provides backpressure, dead-lett…
Anthropic moved the computer use tool out of beta on the Claude API on August 19, introducing a new toolset with batch actions and a configuration system, alongside a new browser toolset for in-browser automation. OpenAI…
Apple Inc. unexpectedly announced new Mac mini and Mac Studio desktops powered by the new M6 and M5 Ultra chips, with the Mac mini's starting price rising to $899 from $799. The M6 is Apple's first chip built on TSMC's 2…
OpenAI has disclosed first measured results for its custom inference chip, Jalapeño, showing a 1.5 to 1.9 times increase in AI work per watt at peak throughput compared to leading commercial systems, and cutting end-to-e…
Nvidia has released the Jetson Orin Nano Super Developer Kit, an upgraded version of its entry-level edge AI computer that boosts AI performance from 40 TOPS to 67 TOPS while maintaining the $249 price. The kit is optimi…
NVIDIA Corp. introduced the Jetson Orin Nano 2, an entry-level edge AI computer that doubles the inference performance of its predecessor, the Jetson Orin Nano Super, delivering 78 trillion operations per second (TOPS) o…
NVIDIA released CUDA Python 1.0 with CUDA 13.3, making Python a supported, first-class way to use the CUDA platform, with components including cuda.core 1.0.0, cuda.compute 1.0.0, cuda.bindings 13.3.0, cuda-pathfinder, a…
NVIDIA announced the Jetson Orin Nano 2, a new robotics computer for entry-level edge AI, delivering 2x the inference performance of its predecessor in the same form factor while consuming 40% less power at the same perf…
A new open-source harness from Ratify Protocol reproduces a cross-company AI agent handoff and its revocation, running offline with `npm run engagement` and verifying all claims. The harness uses Agent Relay's adapter an…
Jalapeño, a new AI inference chip, is setting a benchmark for inference speed by achieving higher compute utilization, power efficiency, and lower latency compared to traditional GPUs, according to the article. The chip'…
Super Micro Computer Inc. shares surged 8% to $37.88 on Tuesday after Taiwanese prosecutors indicted nine individuals, including two former employees of its Taiwan subsidiary, over illegal AI server exports to China, but…
Microsoft announced on August 25, 2026, that it has become an Official Partner of the Toronto Blue Jays, Canada's Major League Baseball team, in a multi-year partnership to explore cloud and AI technologies for fan exper…
Goldman Sachs projects DRAM undersupply of 5.0% in 2026 and 5.9% in 2027, the tightest since 2017, driven by AI server demand. Micron Technology reported fiscal Q3 2026 revenue of $41.46 billion and non-GAAP EPS of $25.1…