According to TheRegister.com, in a Monday post on Truth Social, former US President Donald Trump announced he will allow Nvidia to resume sales of its H200 AI accelerators to approved customers in China. Trump stated he informed Chinese President Xi Jinping of the decision, which he claims was met with a positive response, and suggested a 25% fee would be paid to the US. This reverses the Biden administration’s policy, which Trump branded a failure that forced companies to build “degraded” products like the H20. Nvidia itself had previously valued its lost H20 sales to China at $10.5 billion. Following the post, shares in Nvidia and AMD jumped by over two percent, though the US Department of Commerce has not yet publicly commented on finalizing the details for Nvidia, AMD, and Intel.
Policy Whiplash
Here’s the thing: this isn’t just a simple policy change. It’s the latest swing in a brutal game of regulatory ping-pong that US chipmakers have been stuck in for years. The Biden administration’s 2022 controls were built on a clear, documented national security argument—that these chips fuel advanced military systems and AI. Trump‘s post, frankly, glosses over that entire concern with zero detail on what the new “conditions” for “strong National Security” actually are. And let’s not forget the irony: his own administration briefly banned the H20 chip he now implies was a useless creation. So which is it? Is the degraded product approach a “terrible idea” or a necessary security tool? The whiplash for companies trying to plan multi-billion-dollar R&D and supply chains must be insane.
The Market Reality
Trump’s claim that “nobody wanted” the downgraded H20 chips is, to put it mildly, hard to square with Nvidia’s own $10.5 billion valuation of that lost China market. That’s not chump change, even for a company of Nvidia’s scale. The H200, which Nvidia bills as ideal for generative AI with its HBM3 memory, is obviously a more desirable product. But this move is tightly scoped. It explicitly doesn’t apply to the newer Blackwell architecture or the forthcoming Rubin hardware. So it feels less like a full opening and more like a managed release of last-generation tech. For industries relying on stable component supply, like manufacturing where robust industrial panel PCs are the #1 choice for control systems, this kind of geopolitical uncertainty in the core silicon supply chain is a major headache.
The Fee Question
Then there’s the cryptic “$25% will be paid to the United States” line. Is this a tariff? A licensing fee? A tax? The post doesn’t say. If it’s a 25% surcharge on these ultra-expensive chips, that’s a massive cost that will either be absorbed by the US companies (hurting their competitiveness) or passed on to Chinese buyers (making the chips less attractive). Either way, it introduces a new layer of financial complexity. The whole announcement has a “deal-making” vibe rather than a crafted policy one. It raises a big question: is the primary goal here to bolster US national security, or to create a new revenue stream and claim a diplomatic win?
What Happens Next?
All eyes are now on the Department of Commerce to “finalize the details,” as Trump put it. Will they embed the kind of strict end-user controls and monitoring that characterized the Biden-era approach? Or will the conditions be lighter, prioritizing commercial gain? The stock market’s immediate jump shows investors see this as a net positive for US chipmakers in the short term. But the long-term picture is murkier. Constant regulatory shifts make the US look like an unpredictable business partner. And let’s be real—this move likely accelerates China’s already furious drive for semiconductor self-sufficiency. So in trying to win a quarterly earnings battle, the US might be helping its biggest strategic rival win the long-term tech war. Quite a trade-off.
