AI Infrastructure news and analysis on Web Pulse: 35721 curated articles tracking the latest AI Infrastructure developments, tools, and research, updated continuously from vetted sources.
Apple's Apple Intelligence is a hybrid system that runs privacy-sensitive tasks on-device but routes heavier workloads to Private Cloud Compute (PCC), with offloads including agentic tool use, complex reasoning on AFM 3 …
Apple's macOS 27 'Golden Gate' introduces Siri AI alongside third-party models Claude (Anthropic) and Gemini (Google) as governed enterprise AI integrations, with Apple Business settings separating external intelligence …
Apple is retiring legacy MDM mechanisms across all OS 27 releases with no grace period, forcing a cutover to Declarative Device Management (DDM), a framework where servers declare desired state and devices enforce it loc…
Realtime voice agents, which combine streaming speech-to-text, a large language model, and text-to-speech in a cascaded pipeline, must achieve a time-to-first-audio (TTFA) of around 200 milliseconds to feel natural, with…
Function calling, the Model Context Protocol (MCP), and the Agent-to-Agent (A2A) protocol together decouple AI agents from vendor-specific interfaces, enabling enterprise voice agents to act on backends and coordinate ac…
A technical survey reports that a fully self-hosted voice AI pipeline can achieve around 755 milliseconds from end of speech to first audio output, while a provider comparison shows the same speech-to-text task scores 11…
Groq, an AI chip firm, hired Ola Kasali as head of network and data center engineering after nearly 12 years at Meta, where she rose from lead network engineer in 2017 to manager for network deployment and support in 202…
Alibaba released Qwen3.8-Flash-Next, the first open-weight preview of the Qwen4 architecture, on 2026-08-26, featuring 125B total parameters with only 6B active per token and topping Qwen3.7-Plus-Base on 8 of 14 benchmar…
Z.ai released GLM-5.3-Flash on 2026-08-26, a 320B-parameter mixture-of-experts model with 18B active parameters, the first natively multimodal model in the GLM-5 series and the first open-source frontier model to pair sp…
Amazon is shredding physical books after digitizing them to feed its AI model training pipeline, a process that prioritizes high-fidelity, structured content over messy web data. The industrial workflow involves automate…
A New York Times opinion piece warns that the artificial intelligence investment boom, driven by hyperscalers borrowing heavily to build data centers, is endangering the broader economy, with the debt binge creating syst…
Zhipu AI's GLM-5.3-Flash model demonstrates strong performance in instruction following, context comprehension, and low latency, making it economically viable for large-scale automation, according to a hands-on guide. Th…
Z.ai launched GLM-5.3-Flash, the first natively multimodal model in its GLM-5 family, under an MIT license, featuring 320 billion total parameters and 18 billion active parameters. The company claims it outperforms GLM-5…
Meta Platforms Inc. reassigned approximately 6,500 to 7,000 engineers to its Applied AI unit in spring 2026, triggering a drop in employee favorability from 74% to 55%, and later cut about 8,000 jobs (10% of its workforc…
A developer has proposed that logistics AI systems need memory architecture akin to the Three-Eyed Raven from Game of Thrones, enabling them to retrieve and reason over historical data to inform current decisions. The co…
In Danville, Virginia, local officials are promoting a proposed data-centre campus at Berry Hill as the biggest economic development project in southern Virginia's history, but residents are questioning its environmental…
Google Cloud has integrated native TPU support into vLLM, the open-source LLM serving engine, to enable enterprise-grade precision for long-context multimodal embedding inference, targeting the Qwen3 Embedding model seri…
Lumen Technologies (NYSE: LUMN) expanded its Multi-Cloud Gateway service to connect more than 10 million U.S. business locations to private, high-performance cloud networking, extending beyond its fiber footprint to qual…
George Town Council in Tasmania approved Firmus Technologies' 288MW AI data centre at Long Reach on August 25 with a 6-1 vote, despite 349 written objections and nearly 6,000 petition signatures from locals. The facility…
DeepSeek released an open-source coding agent runtime called Harness on August 13, which amassed 165,000 GitHub stars within a week, including 50,000 in the first twelve hours. The MIT-licensed, Node.js-based runtime, bu…