Nvidia CEO Jensen Huang ignited debate by suggesting China could win the AI race, citing the country’s cheaper, more accessible energy and rapid infrastructure buildout—advantages he later softened by calling China “nanoseconds behind” the U.S. The newsletter argues that if AI leadership hinges on scaling power-hungry data centers rather than marginal algorithmic gains, Beijing’s subsidies, centralized planning, and streamlined approvals provide an edge while U.S. firms face higher costs and grid constraints. China is also surging in open-source usage and cost-efficient innovation, with startups like DeepSeek and research from Tencent showcasing aggressive optimization and fresh techniques. The U.S. retains strengths but is hampered by fragmented regulation and power shortages—Microsoft even reports idle GPUs for lack of electricity. In other AI developments: Anthropic targets break-even years ahead of OpenAI amid divergent spending strategies; Meta’s Yann LeCun is reportedly eyeing a startup; OpenAI faces new lawsuits; and researchers warn advanced reasoning models are easier to jailbreak, underscoring rising safety risks as the race intensifies.
Related articles:
Universal and Transferable Adversarial Attacks on Aligned Language Models
The AI Canon
