AI Infrastructure news and analysis on Web Pulse: 35671 curated articles tracking the latest AI Infrastructure developments, tools, and research, updated continuously from vetted sources.
SpaceX will spend $100 billion to build the largest spaceport on Earth on Pecan Island in Vermilion Parish, Louisiana, a 125,000-acre site named Starbase, Louisiana, that Elon Musk plans to use to launch Starship rockets…
A developer has built HiTranscript, a web app that converts public video URLs and local media uploads into searchable transcripts and subtitle files, and detailed the architecture patterns that ensure reliability. The sy…
Nutanix CEO Rajiv Ramaswami said the company spent $20 million on its own on-premises AI cluster to reduce reliance on external AI coding tools like GitHub Copilot and Anthropic's Claude, and expects to recoup the invest…
Intel's upcoming Datacenter GPU, codenamed Crescent Island, features a 480 GB LPDDR5X memory subsystem, the highest capacity of any GPU or AI accelerator, enabling it to run Deepseek-V4 Flash entirely from VRAM. The GPU,…
Kioxia Holdings plans to build a third chipmaking plant at its site in Iwate Prefecture, Japan, with manufacturing partner Sandisk, to produce high-density 3D flash memory chips for AI workloads. The investment responds …
A Hacker News user proposed an OpenAI-compatible API for open-weight models that guarantees zero prompt retention, no training on user data, and logs only billing metadata. The user, who is considering running it on an R…
MLSentinel, a new free platform for monitoring deployed machine learning models, was announced on Hacker News by developer Narasimha440. The tool tracks model health and alerts users to underlying issues before they esca…
A new GitHub example demonstrates how to store AOTInductor model weights outside the shared library and update them at inference runtime in a thread-safe manner, addressing the lack of documentation for advanced AOTInduc…
Chinese AI lab Z.ai revealed on Wednesday that the mysterious Ox Alpha model was a free preview of its new GLM-5.3-Flash model, which had appeared on OpenRouter and OpenCode on August 20 and became the most popular model…
Anthropic reportedly plans to spend about $45 billion renting AI cloud computing capacity from Nscale for a West Virginia data-center campus, with infrastructure expected to use NVIDIA Vera Rubin chips. The deal signals …
A new benchmark, DataKernelBench, evaluates whether large language models can optimize database queries on GPUs, achieving up to 2.11x speedup over torch.compile on TPC-H SF10 with an H100 GPU. The benchmark translates S…
Researchers propose ExFold, a training-free expert-folding framework that accelerates both prefill and decode phases of Mixture-of-Experts (MoE) inference by projecting excluded expert contributions onto retained experts…
Researchers introduced SHIFT-LLM, a training-free framework that corrects distribution shifts in depth-pruned large language models by inserting Linear Residual Adapters at each pruning site, recovering accuracy without …
Researchers propose Prefix Sliding, a method that discards intermediate reasoning tokens during test-time scaling, capping memory use and enabling existing models to run 3x faster without training while maintaining perfo…
Researchers introduced SAMpLE, an open-source SystemC-AMS framework that integrates machine learning models into virtual prototyping as first-class Timed Dataflow components via a standardized interface. The framework su…
Z.ai's GLM-5.3-Flash model, initially released anonymously on OpenRouter as Ox Alpha, has processed over 20 trillion tokens in its first six days, according to OpenRouter. The model features a mixture-of-experts architec…
Cerebras Systems is demonstrating that its Wafer-Scale Engine (WSE) architecture significantly reduces inference latency for large language models by keeping the entire model on a single silicon wafer, bypassing the memo…
An engineer from Wexa AI benchmarked five graph database platforms, including CognoDB Cloud, on a 352,768-edge citation network under free-tier constraints. The harness measured ingest, traversal, lookup, aggregation, an…
A developer from QVeris has published a practical pattern for giving AI agents access to external APIs via the Model Context Protocol (MCP). The pattern, called 'discover → inspect → probe → call', keeps the agent's init…
A developer detailed the architecture of a dedicated ASP.NET Core guardrail microservice on Kubernetes that validates LLM requests, enables instant policy updates via Redis, and scales with custom HPA for high throughput…