TL;DR

US ties broadband funding to AI deregulation; Nvidia/Nebius set MLPerf records; EleutherAI releases open legal LLM dataset.

Highlights

• U.S. Senate Republicans propose tying $42B in broadband funding to a ban on state-level AI regulation, aiming for uniform federal policy and faster broadband/AI rollout1.
• Trump administration renegotiates CHIPS Act grants with TSMC, Samsung, SK Hynix, seeking lower subsidy ratios; BEAD broadband program is overhauled to remove climate, labor, and rate regulations2.
• Nvidia and Nebius achieve record MLPerf Training v5.0 results with GB200 NVL72 systems; Nebius trains Llama 3.1 (405B params) in 124.5 minutes on 1,024 Hopper GPUs3.
• Nvidia introduces Nemotron-H hybrid model, new inference scaling techniques, and highlights MoE performance improvements with GB200 NVL723.
• EleutherAI, with partners, releases the 8TB/2T-token Common Pile dataset—open, legally vetted, and matching LLaMA 1/2 performance in key tasks4.
• Google tests "Search Live" conversational AI in the Google app, adds "scheduled actions" to Gemini AI Pro/Ultra, and evolves Firebase AI Logic for developers59.
• Anthropic launches a custom AI chatbot for U.S. national security, appoints Richard Fontaine to its Long-Term Benefit Trust, and releases models for classified use6.
• Cursor raises $900M Series C at $9.9B valuation, surpasses $500M ARR, and is adopted by over half of Fortune 500; ClickHouseDB raises $350M for global AI-native analytics7.
• Tesla’s Optimus robot program lead Milan Kovac departs; Ashok Elluswamy (Autopilot AI) takes over as Tesla targets rapid robot deployment8.
• Elon Musk’s xAI seeks $5B in debt financing via Morgan Stanley, faces complications from Trump feud and Stargate UAE exclusion; X partners with Polymarket for prediction markets1012.
• Trump signs executive orders lifting the supersonic flight ban, expanding drone authority, and refocusing federal cybersecurity efforts11.
• Saudi Arabia uses AI and biometrics to reduce Hajj heatstroke cases by 90% for 1.8M+ pilgrims13.
• Volvo unveils AI-adaptive safety belt for the 2026 EX60 SUV, adjusting to occupant and crash conditions14.
• China’s gaokao exam for 13.35M students uses AI surveillance and biometrics for exam integrity15.

Commentary

U.S. federal policy is moving toward centralized AI governance and deregulation, with the Senate’s broadband funding bill and BEAD program overhaul both designed to streamline AI and infrastructure deployment12. The explicit link between broadband funds and a ban on state-level AI regulation signals a push for regulatory uniformity, which could accelerate AI adoption but may also prompt legal disputes with states seeking stricter oversight1.

On the hardware and model training front, Nvidia and Nebius continue to set new benchmarks for large-scale AI training, with the GB200 NVL72 system and novel model architectures like Nemotron-H demonstrating significant gains in efficiency and throughput3. These advances lower barriers for both proprietary and open-source LLM development, as seen with EleutherAI’s Common Pile, which provides a legally sound, high-quality dataset for scalable model training4.

Enterprise AI momentum remains strong: Cursor’s rapid revenue growth and broad Fortune 500 adoption underscore robust demand for AI-native productivity platforms, while ClickHouseDB’s funding reflects continued interest in analytics infrastructure7. Google is expanding both consumer-facing and developer AI tools, pushing further into proactive assistant capabilities and hybrid LLM deployment59.

Anthropic’s deployment of custom AI for U.S. national security and Tesla’s leadership transition in the Optimus program highlight the growing integration of AI into both government and industrial operations68. Meanwhile, xAI’s financing challenges and X’s Polymarket partnership illustrate the complex intersection of AI, capital markets, and regulatory risk1012.

AI’s operational role in public safety and compliance is broadening, as demonstrated by Saudi Arabia’s Hajj management, Volvo’s adaptive safety systems, and China’s AI-powered exam surveillance131415. These cases reinforce the importance of legal, ethical, and technical safeguards in large-scale deployments. AI professionals should monitor U.S. regulatory developments, hardware advances, and the evolving funding climate for both opportunities and compliance risks.