I suppose you could say I’m sympathetic to export controls aimed at curbing Xi Jinping’s access to advanced chips.

We toss around the word “dictator” pretty freely these days, but it applies in a very strict sense to Xi. China’s not one-party rule anymore. It’s one-man rule, and Xi’s not to be trusted. Democracy, being famously messy and often inefficient, is anathema to his constitution. He doesn’t believe in it, nor in the principles which inform it and stem from it.

To the extent AI’s the key to the future, and he who wields the most powerful AI could claim a kind of dominion over humanity, Western democracies have a good case for keeping the best chips out of Xi’s hands.

That said, there’s also a case to be made that the harder you try to keep China from obtaining something, the harder they’ll try to produce it internally. Being a smart people, they’ll figure it out eventually, whatever “it” is. Semis and AI are no exception.

Xi’s almost pathological aversion to capitalism serves as an impediment — the incentive structures which facilitate innovation in the US aren’t there in China — but he can compensate for that, at least in part, with state-led investment. That’s something to keep in mind when you read about Donald Trump’s budget cuts.

It’s not completely lost on the administration that in their rush to cut federal jobs and spending, they might inadvertently choke off crucial R&D. Michael Kratsios, Trump’s science and technology director, said this week that scarce federal money for research needs to be disbursed “rapidly,” but his remarks, at an event in Austin, alluded to the challenging environment. The US, he said, “must be more creative in our use of public research and development money.”

Earlier this year, the National Science Foundation cut nearly 200 jobs, or more than 10% of its workforce. “You screwed people,” an all-staff email signed by affected employees read. “Take some accountability.” One employee accused the agency of taking scalps to present to the Office of Personnel Management. A spokesman for NSF said, at the time, that the agency was just trying to comply with Elon Musk’s DOGE.

Democrats on the House Science, Space and Technology Committee weren’t amused, calling the cuts “a disgrace and international embarrassment.” “Indiscriminately firing 10% of NSF’s workforce is a massive mistake that will hurt our ability to compete globally, especially with China,” Haley Stevens and Zoe Lofgren seethed, adding that America’s “science and innovation enterprise will no longer be the envy of the world if Co-Presidents Elon Musk and Donald Trump get their way.” Musk and his “minions,” they went on, in the same statement, have “no idea what they’re doing.”

For reference, the NSF was created three quarters of a century ago with a mandate to preserve and perpetuate America’s edge in science and engineering. 25 cents of every dollar in “basic” federal research grants to colleges and universities goes through the agency. And wouldn’t you know it, there’s a link between that research and AI development in America.

As one expert who spoke to Bloomberg sighed, “almost every employee with an advanced degree at every American AI firm has been a part of NSF-funded research at some point in their career.” The same linked article noted that the foundational work behind Google’s search algorithm was made possible by an NSF grant.

That’s why you can’t put people like Trump in charge: They’re likely to do very silly things which undercut their own agenda and, more importantly, undermine the future of the country, or even humanity more broadly. In this case, that means choking off R&D critical to tomorrow’s technology at a time when America’s strategic adversary, in China, is spending more and more on the very same R&D.

The biggest market story on Wednesday was Nvidia’s announcement that Trump’s now requiring a license to ship the H20 chip it designed specifically for the Chinese market, a decision that’s expected to cost the company $5.5 billion this quarter. AMD, in its own filing, said that the same export controls could cost the company “up to approximately $800 million in inventory, purchase commitments and related reserves.”

Coming full circle, I think those measures are defensible considering what it is, and who it is, we’re talking about (AI and Xi, respectively). But if you’re going to go the export control route, don’t turn around and undercut the whole effort by curbing federal R&D funding, particularly not in slapdash fashion. Beijing’s going to see that as an opportunity. And they won’t be wrong.

Insult to injury Wednesday in the chip space was a lackluster orders readout from ASML which, in the press release accompanying its quarterly results, lamented a lack of operating visibility. “Our conversations so far with customers support our expectation that 2025 and 2026 will be growth years [but] the recent tariff announcements have increased uncertainty in the macro environment and the situation will remain dynamic for a while,” the company said.