Q&A with Google Cloud CEO Thomas Kurian on building infrastructure for AI agents, balancing internal needs and the demands of customers like Anthropic, and more

Nvidia CEO Jensen Huang gave a rare extended interview to Stratechery's Ben Thompson, covering a sprawling set of topics central to the GPU giant's future.

Huang discussed CUDA — the software layer that's arguably as important as Nvidia's hardware dominance — and how reasoning and coding workloads are reshaping demand for accelerated computing. He addressed the evolving role of CPUs alongside GPUs, pushing back on the idea that traditional processors are irrelevant in the AI era.

The conversation touched on Groq, the AI chip startup positioning itself as a competitor with its LPU architecture. Huang also weighed in on China's access to advanced chips and dismissed AI doomsday narratives.

Perhaps most revealing: Huang reflected on Nvidia's fundamental nature as a company — not just a chipmaker, but a full-stack accelerated computing platform. The interview signals how Nvidia sees the next phase of AI infrastructure playing out.