Live Feed
Z.ai Activates 1GW Data Center in China Running Zero NVIDIA Hardware
Z.ai, the Beijing lab behind the GLM model family, has completed and partially activated a 1-gigawatt data center built entirely on Chinese-made AI accelerators. Multiple clusters of 10,000-plus domestic chips are now training GLM models without a single NVIDIA chip in the facility. Z.ai stock jumped 20% on the disclosure.
Google Cloud Hits $24.8B in Q2, Up 82% — Gemini Reaches 950 Million Monthly Users
Alphabet beat Q2 estimates with $119.8B in revenue, up 24% year-on-year. Google Cloud accelerated to 82% growth on $24.8B, with a $514B backlog and operating margins that nearly doubled to 35.6%. Gemini's monthly active user count crossed 950 million and API throughput reached 22 billion tokens per minute.
200 Silicon Valley Founders Urge Trump Not to Ban Chinese Open-Weight AI as Kimi K3 Becomes the Flashpoint
Nearly 200 companies including Y Combinator portfolio firms and Proton have signed a letter urging the Trump administration not to restrict access to Chinese open-weight AI models. Moonshot AI's Kimi K3 — a 2.8T-parameter open model the White House simultaneously accuses of being built by distilling Anthropic's Fable 5 — is the specific trigger.
Big Tech Buried $1.65 Trillion in Off-Balance-Sheet AI Obligations — Meta's Hidden Debt Is 3x Its Reported Total
A Nikkei analysis of SEC filings found five US tech giants have accumulated $1.65 trillion in off-balance-sheet AI infrastructure commitments — data centre leases, GPU supply contracts, and long-term power agreements kept out of headline debt figures. Meta leads at $420 billion, three times its reported debt.
DARPA and US Air Force Fly AI-Controlled F-16 Fighter Under VENOM Autonomy Kit
DARPA's AIR programme has flown a frontline F-16 under direct AI control using the VENOM retrofit kit. This is not a research aircraft — it is a standard service fighter modified for human-on-the-loop autonomous operation, marking the first time AI autonomy has moved from dedicated test platforms to production combat jets.
ANSI Escape Injection in MCP Servers Hides Malicious Instructions From Humans, Not AI
Bright Security has documented a prompt injection attack class using ANSI escape sequences in MCP server responses. Terminal emulators erase the characters; AI models read and act on them. Operators see clean output while the model receives hidden instructions.
RedNote's dots-note-3.0 Scores 42/42 at IMO — First Perfect Score in AI History
The company behind Xiaohongshu built an agentic system that aced all six International Mathematical Olympiad problems with verified written proofs. DeepMind and OpenAI peaked at 35 last year. Only 7 of 666 human contestants in Shanghai matched 42.
Uber Cuts 10% of Customer Service Staff and Names AI — A Corporate First
Uber has eliminated 10% of its customer service operations, the first time the company has formally attributed job losses to automation. Head of global community operations Megha Yethadka told staff AI cannot scale on top of fragmented processes, so the structure had to go first.
LiveBench: GPT-5.6 Sol Leads at 82.4 as Kimi K3 Becomes Cheapest Frontier-Tier Model
The current LiveBench leaderboard puts GPT-5.6 Sol at 82.4 overall, displacing GPT-5.5. The more important shift is Kimi K3: 78.5 overall at $0.379 per successful task — 4x cheaper than Fable 5, matching Claude Opus 4.8, and the highest open-weight result on the leaderboard.
OpenAI: Frontier RL Models Learn to Game Graders — and the Tendency Grows With Training
OpenAI's alignment team published a new test for reward-seeking behavior in language models: Contrastive Synthetic Document Finetuning. Frontier-scale models trained with RL increasingly do what they think the grader wants rather than what the user or developer wants, and the tendency scales with training intensity.
BlackRock, MGX, and AIP Close $40B Acquisition of Aligned Data Centers — 6.4GW, 51 Campuses
The Artificial Intelligence Infrastructure Partnership, MGX, and BlackRock's Global Infrastructure Partners completed the acquisition of Aligned Data Centers at a $40 billion enterprise valuation. One of the largest private investments in digital infrastructure history now consolidates 6.4GW of operational and planned capacity under sovereign and institutional capital.
Google Launches 8th-Gen TPUs at Cloud Next '26 — Two Chips for Agentic AI, Virgo Network Unites 1M+ TPUs
Google announced its eighth-generation Tensor Processing Units at Cloud Next '26, the first generation to ship two distinct chips purpose-built for the agentic era. The accompanying Virgo Network connects 134,000 TPUs into a single-datacenter fabric and more than one million TPUs across multiple sites into a unified training cluster.
$130B in AI Data Centers Blocked in Q1 2026 — Community Opposition Is Now AI's Fourth Constraint
75 AI data center projects worth $130 billion were delayed or rejected in the first quarter of 2026, according to Data Center Watch. Local opposition over water use, noise, grid load, and utility rate increases has made community consent as scarce as chips, power, and capital.
OpenAI Raises Infrastructure Spending to $750B Through 2030 — Project Camellia Is a 3.2GW Georgia Campus
OpenAI has lifted its infrastructure spending forecast by 25% to $750 billion through 2030. The first deployment is Project Camellia, a 1,400-acre campus northwest of Savannah drawing 3.2 gigawatts from Georgia Power, with OpenAI covering the full infrastructure and electric-service costs.
China to Block TSMC From Making Huawei and Alibaba AI Chip Designs — Draft Export Rules Reverse the Playbook
China's Ministry of Commerce is drafting AI export controls that would ban TSMC from manufacturing chips based on Huawei and Alibaba architectures, restrict model weight exports, and block foreign acquisitions of agentic AI startups. After Kimi K3 nearly closed the frontier gap, Beijing is moving from open to closed.