AI Infrastructure news and analysis on Web Pulse: 34521 curated articles tracking the latest AI Infrastructure developments, tools, and research, updated continuously from vetted sources.
A developer's analysis of LLM agent orchestration found that idle parent agents on Codex burn tokens polling for worker completion because the runtime lacks event-driven wake-ups, with one run issuing 89 waits that consu…
Meta Vice President of Infrastructure Manish Kadakia said at the AI Infra Summit in Santa Clara on Sept. 16, 2026, that all four core data center systems — servers, networks, electricity and cooling — are changing simult…
A developer worked around DynamoDB's TopK=100 cap on vector search results by sharding the vector index partition key into five hash-derived groups and querying them in parallel, then merging and deduplicating the per-sh…
A study measuring 54 configurations of Qwen2.5-7B-Instruct on vLLM 0.12 across L4, A100, and H100 GPUs found that 18 of 36 configurations reach the cost, quality, and latency Pareto frontier, with combined optimization m…
A training-free routing policy called the Tri-Metric Router achieved 0% out-of-memory failures and 88.5 ± 4.4% oracle alignment on out-of-distribution holdouts for retrieval-augmented generation on commodity GPUs, accord…
Researchers introduced Fathom, a per-query read-depth key scan for sparse decoding over offloaded KV caches, which at one million tokens on Qwen3-8B runs a decode step 1.67x faster in GPU time than the 136-bit scans of D…
Researchers released Zing-0.5, a 5B autoregressive world model that supports joint keyboard and online text control for real-time playable generated worlds, according to the arXiv paper 2609.17909v1. Zing-0.5 scores 81.0…
EffiRAG, a graph-based retrieval-augmented generation system, produced the preferred answer on 93 of 120 open-ended questions in the UltraDomain benchmark while cutting total system cost by 57 percent, from USD 0.952 to …
Researchers proposed WFM, a Wiki Foundation Model for scalable, agent-native knowledge representation and retrieval, detailed in arXiv paper 2609.18182v1. WFM formalizes a Wiki Graph schema coupling dense document contex…
Researchers introduced AutoTuneBench, a benchmark and measurement protocol that makes trust architectural for LLM agents auto-tuning GPU kernels and serving engines, after characterizing four failure modes from a four-da…
A security measurement found 33,005 internet-reachable LiteLLM gateway instances following CVE-2026-59822, an authentication bypass in LiteLLM's MCP Streamable HTTP endpoint that was fixed in version 1.84.0 and added to …
Huawei Technologies will launch two new artificial-intelligence semiconductors in 2027, rotating chairman David Wang said on Thursday, with the 960DT due in the first quarter and the Ascend 960PR in the third quarter as …
Researchers introduced VC-Attention, a method combining value smoothing and softmax casting to enable low-bit attention for Diffusion Transformers used in video generation. The work targets the dominant deployment cost o…
A developer outlines a framework for selecting multimodal AI APIs by matching platform architecture to workload type, splitting options into four categories: cross-provider gateways, Replicate for experimentation and cus…
A developer built a suite of twelve tag-scoped MCP tools to inventory and manage a single AMD Instinct MI300X GPU on AMD Developer Cloud, using the same server to read the card's native numeric format support directly of…
Texas Governor Greg Abbott has reversed course on the AI data center tax subsidies and deregulation he championed, now demanding an end to the giveaways and calling for regulation of projects he previously shielded from …
Huawei Technologies unveiled its Ascend 960 SuperPoD computing cluster and an upgraded UnifiedBus interconnect technology at the Huawei Connect 2026 conference in Shanghai on Thursday, saying the combination of UnifiedBu…
A new open-source tool called Overlord provides a trust kernel for AI agents, giving them transactional isolation, provenance, and arbitration for untrusted execution in dependency-free Python on Linux kernel primitives.…
An enterprise engineering team's stress test of an automated reconciliation pipeline found that multi-step LLM agent loops collapse under sequential reliability math, with roughly 88% of autonomous multi-step agent proje…
A developer built Baize, a lightweight AI agent runtime written in Go that runs as a sidecar process alongside existing services, converting OpenAPI documentation into callable tools and gating important writes behind hu…