Jensen Huang (黄仁勋) reportedly has no successor, says he “hopes to die suddenly at work” and that AI may already be AGI

Leader: succession by diffusion, not by heir

Jensen Huang (黄仁勋), the long-time founder and CEO of NVIDIA, reportedly told a recent 145‑minute podcast that he does not believe in a traditional succession plan and has instead spent years pouring his mental model into the company. It has been reported that he even said he “hopes to die suddenly at work” rather than leave behind a knowledge gap — because, he argues, he has tried to make his reasoning visible to dozens of direct reports so the company can continue without a single replacement. Is that confidence in institutionalized knowledge reassuring — or a risky single‑point‑of-failure disguised as decentralization?

The technical thesis: AGI, agent scaling, and trillion‑dollar math

On the same podcast Huang reportedly made two headline-grabbing claims: “I think we have achieved AGI,” he said, with the caveat that the answer depends on how one defines AGI. More consequential for markets was his breakdown of future compute demand: pre‑training, fine‑tuning, inference, and what he calls “agent scaling.” Huang argued that once AI systems routinely decompose tasks into cooperating agents, inference becomes continuous, stateful and massively parallel — more like running a factory every time a user asks something than like a single cached lookup. That, he said, is why he believes demand for high‑end compute could scale orders of magnitude beyond today’s levels and why NVIDIA could, in theory, aim for revenue figures that today seem staggering.

Geopolitics, China and the human element

Huang also praised China’s rapid innovation speed and credited dense social networks — overlapping school, hometown and industry ties — with accelerating knowledge flow there. For Western readers: this matters because the U.S. and China are already locked in a tech rivalry shaped by export controls and sanctions on high‑end chips; NVIDIA sits at the center of that squeeze, supplying both hardware and the software stacks that power generative AI. Whether one agrees with his AGI claim or his $3‑trillion revenue thought experiment, Huang’s broader point landed on a human register: his role, he said, is less CEO than a “knowledge router,” and the company’s future depends on how densely that knowledge is shared across people, not on any single successor.