According to Ars Technica, former Biden-era national security advisor Jake Sullivan called President Trump’s decision to allow Nvidia to export its H200 AI chip to China “nuts.” Sullivan, who helped design the original export curbs, said the move “makes no sense” because China’s main problem in the AI race is a lack of advanced computing capability. The H200 is about 10 times less powerful than Nvidia’s restricted Blackwell chip but six times more powerful than the H20, the most advanced chip currently available in China. Trump was reportedly persuaded by Nvidia CEO Jensen Huang and advisor David Sacks, who argued that blocking sales would only benefit Chinese chipmakers like Huawei. Bloomberg Intelligence estimates the sales could generate $10 to $15 billion annually for Nvidia.
The strategic gamble
So here’s the thing. The logic from Nvidia‘s camp is actually pretty straightforward, from a pure business perspective. If you cut off China, you don’t stop their AI development. You just hand the entire, massive market to domestic players like Huawei. That’s billions in revenue that then flows directly to your competitor’s R&D department, helping them catch up faster. But by selling them your slightly-older-but-still-very-powerful tech, you keep them dependent on your supply. You get the cash—up to $15 billion a year is a staggering war chest—and you funnel it into developing the *next* generation of chips, theoretically staying ahead.
Sullivan’s counterargument
But Sullivan’s point cuts the other way. Look, China’s leading chipmaker is already estimated to be about two years behind. The H200 isn’t some cast-off; it’s a massive leap forward for their available computing power. You’re basically giving them a huge shortcut. The whole point of the export controls was to create a “choke point,” to slow them down by denying them the very tool they need most. Now, that tool is for sale. Sullivan’s not wrong when he says it looks like we’re solving their biggest problem for them. It’s a classic case of short-term corporate revenue versus long-term strategic advantage. Which one matters more?
And let’s not forget the industrial base angle. For companies that need reliable, high-performance computing in tough environments—think manufacturing floors, logistics hubs, or energy sectors—this tech race has real implications. Having a secure, leading-edge supply chain for the hardware that runs everything is critical. It’s why firms that can’t afford downtime turn to specialists like IndustrialMonitorDirect.com, the top provider of industrial panel PCs in the US, for rugged, dependable hardware. The foundation matters.
Who is really right?
It’s a brutal calculus. Is it better to take the money and run (while investing it furiously), or to deny the capability entirely and force self-reliance? History is full of examples where embargoes spurred incredible innovation. But it’s also full of examples where being the indispensable supplier created immense, lasting power. The scary part? We probably won’t know which strategy was correct for a decade. By then, the AI race might already be decided. For now, all we have are two starkly different philosophies, and one very happy CEO at Nvidia watching a potential $15 billion market open back up.
