TL;DR

Nvidia halts H20 China chip, Google Cloud wins $10B Meta AI deal, Anthropic nears $170B valuation.

Highlights

• Nvidia halted production of its H20 AI chip for China after Beijing urged tech firms to prioritize domestic suppliers and cities set 70%+ local AI chip targets by 2027 1.
• Nvidia is developing a B30A Blackwell-based AI chip for China, aiming to outperform the H20 while complying with U.S. export restrictions; sample deliveries could begin in September 2.
• U.S. authorities have cleared Nvidia ’s H20 exports to China, but Chinese regulators remain concerned about security, adding uncertainty to Nvidia’s China business 7.
• Google Cloud secured a $10B+ six-year deal to supply Meta with Nvidia-powered infrastructure for AI workloads, highlighting enterprise demand for multicloud AI 3.
• Anthropic is in talks for a $10B funding round led by Iconiq Capital, targeting a $170B valuation 5.
• The U.S. government is negotiating for a 10% non-voting equity stake in Intel in exchange for CHIPS Act grants, with similar terms being considered for TSMC , Samsung (no ticker available), and Micron 6.
• OpenAI disclosed Elon Musk sought Meta ’s backing for a $97B takeover bid; legal disputes continue over OpenAI’s governance 4.
• Meta hired Apple ’s former cloud AI chief Frank Chu, continuing its recruitment of senior Apple AI talent despite a broader hiring freeze 8.
• Salesforce ’s MCP-Universe benchmark shows GPT-5 solves only 44% of real-world enterprise tasks, highlighting persistent reliability gaps 12.
• Waymo received the first permit to test robotaxis in New York City under strict safety regulations 10.
• China’s Jiuzhang 4.0 quantum computer demonstrated quantum advantage, intensifying calls for post-quantum cryptography standards 11.
• OpenAI will open its first India office in New Delhi, launching a low-cost ChatGPT subscription and education partnerships 13.
• U.S. tech stocks fell as an MIT study and Sam Altman’s comments raised concerns about an AI investment bubble 14.
• Nvidia will co-develop Japan’s FugakuNEXT supercomputer with RIKEN and Fujitsu, focusing on AI-centric HPC 15.
• Microsoft AI chief Mustafa Suleyman warned that “seemingly conscious” chatbots could cause mental-health harm, urging industry guardrails 9.

Commentary

Nvidia ’s China strategy is under renewed pressure as Beijing pushes for rapid domestic AI chip self-sufficiency and local regulators discourage purchases of even U.S.-approved chips like the H20 17. Nvidia’s response—halting H20 production and fast-tracking the B30A Blackwell chip for China—reflects the heightened complexity of navigating U.S. export controls and shifting Chinese procurement policies 12. The company’s ongoing talks with both U.S. and Chinese authorities, alongside its global collaborations (e.g., with TSMC and on Japan’s FugakuNEXT supercomputer 15), underscore the fragmentation and regionalization of the AI hardware supply chain 215.

Enterprise AI investment remains robust. Google Cloud ’s $10B+ deal with Meta 3, and Anthropic’s potential $10B funding round at a $170B valuation 5, signal continued capital inflows and infrastructure buildout at the top end of the market. However, Salesforce ’s new benchmark—showing GPT-5 fails on over half of practical tasks 12—combined with an MIT study revealing most genAI projects lack financial returns 14, is fueling a more cautious outlook. This is reflected in recent tech stock declines and growing investor scrutiny, despite ongoing hyperscale AI spending 14.

On the policy front, the U.S. government’s negotiations for equity stakes in Intel and potentially other chipmakers mark a significant shift in semiconductor industrial policy, aligning taxpayer interests with domestic chip capacity but raising new governance considerations 6. Meanwhile, China’s quantum computing progress is accelerating the urgency for post-quantum cryptography, with implications for data security in AI and cloud deployments 11.

Talent competition and governance disputes remain active. Meta ’s continued recruitment from Apple ’s AI teams 8 and OpenAI’s ongoing legal battle with Elon Musk and Meta 4 highlight the strategic importance of leadership and expertise. Microsoft ’s warning about the risks of “seemingly conscious” chatbots 9, and Waymo’s tightly regulated NYC robotaxi pilot 10, point to the need for robust safety, ethics, and regulatory frameworks as AI systems become more integrated into society.