AI Infrastructure

Perplexity AI announced the first hybrid local-server inference orchestrator at Computex 2026, automatically routing AI tasks between a user's device and cloud-based frontier models based on data sensitivity and compute …

NVIDIA at GTC 2026 presented the full-stack confidential computing capabilities of the Vera Rubin architecture, emphasizing that future AI chips must keep data encrypted throughout computation to prevent access by cloud …

NVIDIA and Apple have resolved the hardware bottleneck for on-device AI, with NVIDIA's RTX Spark (Blackwell GPU, Grace CPU, 128GB unified memory) and Apple's M-series chips enabling 4B+ parameter models to run locally. T…

A developer building a multi-repo personal code factory argues that while AI coding is "solved" through a combination of capable models, action harnesses, deterministic constraints, and project-specific skills, the archi…

A developer built a real-time activity logging and security scoring dashboard for Claude Code that tracks every file read, command executed, and API call with risk labels and timestamps. The tool, available as an npm pac…

Willow, an Israeli startup founded by three former Wix engineers, raised $7 million in a seed round led by Hetz Ventures to build an identity and access management platform for autonomous AI agents. The company’s platfor…

Microsoft released a tutorial demonstrating how to run a browser-use agent workflow using Microsoft Fara in Google Colab. The tutorial guides users through setting up a mock OpenAI-compatible endpoint to test browser act…

HUD AI (YC W25) is offering developers up to $10,000 to license their old, private codebases as training data for Frontier AI labs. The company has already facilitated millions of dollars in transactions over the past mo…

42% of companies are now considering moving AI workloads back on-premises, driven by data sovereignty, intellectual property protection, and cost concerns. For regulated industries like finance, healthcare, and governmen…

Agentic AI systems, which autonomously plan, execute tasks, and iterate on results, can consume hundreds of thousands to millions of tokens per day, with annual token costs per agent ranging from $1,095 to $3,833. Howeve…

Teams building AI applications are discovering that embedding pipelines for retrieval-augmented generation (RAG) systems are fundamentally a data engineering problem, not a new AI discipline. The process of ingesting, ch…

The InfoQ article series "Securing the AI Stack: From Model to Production" examines how AI's shift from experimentation to production has created a volatile security landscape defined by data poisoning, AI-driven phishin…

Google released LiteRT-LM, a new runtime framework that accelerates on-device inference for its Gemma 4 large language models by up to 2.2x using multi-token prediction. The framework, built on the LiteRT platform former…

Intel showcased six new thin-and-light laptop designs powered by the Intel Core Series 3 processor family at Computex 2026, offering up to 20 hours of battery life, built-in AI acceleration, and Thunderbolt 4, Wi-Fi 7, a…

Major tech firms and AI labs are pulling back on internal AI usage after rapid adoption drove unexpected cost spikes, with Uber burning through its entire 2026 AI coding-tools budget in four months, according to Fortune.…

Nvidia CEO Jensen Huang said robotics "is going to be the next major sector" in South Korea after arriving at Gimpo International Airport, with meetings scheduled at Hyundai Motor, LG, SK Hynix, Samsung Electronics and N…

Nvidia CEO Jensen Huang arrived in Seoul on a four-day visit Monday, declaring robotics and physical AI as Korea's next major growth sector. The visit aims to expand Nvidia's relationship with Korea beyond the high-bandw…

Mochallama, a new Java library, enables running llama.cpp inference directly within a Java process using JDK 22's Foreign Function and Memory (FFM) API, eliminating the need for separate daemon processes or native builds…

Monterey Park, California, voters approved Measure NDC with 86.27% support on June 2, enacting the first citywide permanent ban on data center development in the United States. The prohibition, triggered by public opposi…

NVIDIA GPUs handle variable-length inference requests dynamically without recompilation, while TPUs and AWS Trainium require fixed shapes compiled ahead of time, causing crashes or stalls on mismatched inputs. This archi…