Jensen Huang – TPU competition, why we should sell chips to China, & Nvidia’s supply chain moat

Jensen Huang – TPU competition, why we should sell chips to China, & Nvidia’s supply chain moat

More

Descriptions:

Dwarkesh Patel sits down with Nvidia CEO Jensen Huang for one of the most substantive interviews the company’s founder has given on Nvidia’s strategic position, supply chain commitments, and long-term philosophy. Huang addresses the core question of whether Nvidia is vulnerable to commoditization head-on: he argues that the transformation of electrons into valuable tokens requires deep, hard-to-replicate engineering across the full stack—silicon, interconnects, libraries, and ecosystem—and that this journey is far from over.

On supply chain, Huang confirms and contextualizes the reported $100–250 billion in purchase commitments with TSMC, SK Hynix, Micron, Samsung, and ODMs, explaining that much of the upstream investment is implicit—suppliers commit capital because Huang has personally convinced their CEOs of the market’s scale. He also confirms Nvidia is backstopping CoreWeave to $6.3 billion with a $2 billion direct investment, defending this as consistent with his philosophy of doing ‘as much as needed, as little as possible’—enabling the neocloud ecosystem to exist without Nvidia becoming a hyperscaler itself.

Huang makes the case for selling chips to China, arguing that restricting access pushes the world off the American technology stack. He also predicts that AI agents will cause software tool usage to skyrocket rather than collapse incumbent vendors, using Synopsys and Cadence design tools as examples of products that will see agent-driven demand surges. The conversation spans CUDA’s 20-year investment history, domain-specific libraries like cuLitho, and TPU competition from Google and others.


📺 Source: Dwarkesh Patel · Published April 15, 2026
🏷️ Format: Interview

1 Item

Channels

3 Items

Companies

1 Item

People