The U.S. is easing restrictions to allow Nvidia and AMD to sell advanced H200 AI chips to China under strict conditions, potentially unlocking a $50 billion market for Nvidia but with ongoing supply and regulatory challenges. The episode also highlights major funding in robotics and AI security startups, shifting bottlenecks in AI infrastructure, and key developments at Netflix, Spotify, Airbnb, and in U.S. tech legislation.
The U.S. Commerce Department is moving closer to allowing Nvidia and AMD to sell their advanced H200 AI chips to China, marking a significant shift in export policy. Previously, the U.S. operated under a “presumption of denial” for such exports, effectively banning them. The new approach will review export license requests on a case-by-case basis, with strict requirements: companies must prove exports won’t create domestic shortages, that manufacturing for China won’t displace U.S. production, and that robust “know your customer” procedures are in place to prevent unauthorized use. However, the policy limits exports to no more than 50% of U.S. production, and it remains unclear how this cap will be calculated or enforced.
Nvidia’s potential revenue from China is substantial, with CEO Jensen Huang estimating a $50 billion annual opportunity. Demand from Chinese tech giants like Alibaba and ByteDance is high, but supply constraints and regulatory uncertainty persist. Analyst Beth Kindig argues that Wall Street estimates for Nvidia’s future revenue are likely too low, especially if Chinese sales resume. She also notes that despite China’s efforts to develop domestic AI chips, Nvidia’s technological lead remains formidable, making it difficult for competitors—either in China or the U.S.—to catch up in the near term.
The discussion also touches on broader supply chain challenges, particularly the surging price of copper, which is essential for data center expansion and the electric grid powering AI infrastructure. Kindig emphasizes that the bottlenecks in AI are shifting from GPU availability to power and networking constraints. Companies like AMD and Intel are strong contenders in the CPU market, but the overall ecosystem—including memory, bandwidth, and energy supply—will be critical for supporting the next generation of AI systems.
Elsewhere in tech, the episode covers major funding rounds for robotics and AI security startups. Skild AI, a robotics company, raised $1.4 billion to develop a universal “robotic brain” that can adapt across different environments and tasks, focusing initially on enterprise applications like delivery, security, and manufacturing. Meanwhile, Depth First, an AI-powered cybersecurity startup, secured $40 million to build systems that can proactively identify and fix software vulnerabilities, reflecting the growing importance of AI in both offensive and defensive security measures.
Other notable stories include Netflix revising its bid for Warner Bros. Discovery to an all-cash offer to outmaneuver rival Paramount-Skydance, and Spotify’s new co-CEOs facing challenges from AI, user fatigue, and competition from YouTube. Airbnb is also pivoting to integrate more AI features under a new CTO from Meta, aiming to personalize user experiences and boost growth. The episode concludes with updates on legislative efforts to address nonconsensual AI-generated images, with the U.S. Senate passing a bill to allow victims to sue creators of such content, and ongoing Supreme Court deliberations on President Trump’s tariffs.