Analysis
Nvidia CEO Jensen Huang pushed back directly on the market panic surrounding Moonshot AI's Kimi K3 model, telling Axios in an exclusive interview that Wall Street has "misunderstood the impact of Kimi" and that the selloff dragging chip stocks into a bear market misreads what cheap, capable open-weight models actually do to compute demand.
Huang's argument inverts the thesis that's been driving the last week of trading: rather than cheap Chinese models undercutting the case for expensive chips, he argues free and low-cost AI expands the total number of people and businesses using AI at all, and that expansion is unambiguously good for the hardware underneath it. "Free AI should be great for hardware," Huang said. "Free AI should be great for chips. Free AI should be great for data centers." More users experimenting with AI, in his framing, eventually becomes more inference workload running somewhere -- and that somewhere still needs Nvidia silicon.
He also directly rejected the security framing that's followed Kimi K3 and other Chinese open-weight releases -- the idea that running a downloaded Chinese model creates some kind of backdoor to Beijing. Huang argued enterprises can run these models inside sandboxes they fully control and customize, making the security risk a matter of deployment discipline rather than an inherent property of the model's origin.
Most pointedly, Huang told Axios there is "zero possibility" of a scenario where Chinese AI labs run American companies out of the race entirely -- a direct, on-the-record rebuttal to the market narrative that's wiped roughly $3.3 trillion off chip-sector value since late June.
The comments matter because of who's making them: Huang has the widest, most granular view of real AI compute demand of anyone in the industry, sitting at the center of order books for every major lab, hyperscaler and now sovereign buyer. When he says the market has the causality backwards, that's a data point that carries real weight against a thesis that's currently driving billions in daily trading volume.
What to watch: whether Nvidia's next earnings call shows order books holding up despite the Kimi K3-driven panic, and whether other chip executives echo or contradict Huang's more-is-more framing of open-model proliferation.
