Trump's Nvidia Deal May Be a Political Win That Never Pays Off

Jensen Huang walked out of the White House last week with something more valuable than a handshake photo. He had the president's ear on export controls that could unlock billions in Chinese revenue. By Monday, Trump announced on Truth Social that Nvidia could ship H200 chips to "approved customers" in China, with 25% of sales flowing to the U.S. government.

Traders pushed Nvidia shares up 2% in after-hours trading. A lobbying triumph for the world's most valuable company. And it may amount to nothing.

Beijing discouraged Chinese companies from purchasing Nvidia's H20 chip earlier this year, citing alleged "backdoor security risks." The H200 is more powerful, which makes it more strategic but also makes Beijing more likely to block purchases as part of its campaign to wean Chinese industry off American technology. Trump secured permission to sell chips that China's government may not allow its companies to buy.

The 25% fee mechanism appears legally dubious, and the strategic bet relies on assumptions about Chinese behavior that recent evidence contradicts. Trump is already running his victory lap. The checks may never clear.

The 25% Question Nobody Can Answer

Trump's announcement contained a detail that made trade lawyers blink. "25% will be paid to the United States of America," he wrote. Government lawyers are scrambling to find a law that lets them do it. U.S. code prohibits charging fees for export licenses.

The administration's solution involves creative routing. According to Commerce Department officials, the 25% would be collected as an import tax when chips travel from manufacturing sites in Taiwan to the United States for security review. After inspection by the Bureau of Industry and Security, the chips would ship to Chinese customers.

This mechanism transforms an export license into a tariff structure. Chips fabricated in Taiwan, designed in California, bound for Beijing, would generate revenue for the U.S. Treasury during a brief American layover. Whether this survives legal challenge remains uncertain.

The fee itself jumped from 15% to 25% since August, when Nvidia and AMD agreed to share revenue from H20 sales. That arrangement never generated payments because the regulations never got written. The same void could swallow this deal.

Huang told reporters in October that government officials were hammering out a new policy to collect the fees. That was four months ago. The policy still doesn't exist.

Commercial stakes are substantial. Nvidia's CFO Colette Kress said in August that opening trade could mean $2 billion to $5 billion in quarterly China shipments. Huang himself has called China a $50 billion market opportunity. At 25%, successful implementation would route $500 million to $1.25 billion per quarter to Treasury.

If the fees ever materialize.

Beijing's Veto Power

The more fundamental obstacle sits in Zhongnanhai, not Washington. Chinese authorities have demonstrated willingness to reject American chip offerings that don't meet their strategic requirements.

When the Trump administration approved H20 exports in August, China's cybersecurity regulator summoned Nvidia to explain alleged backdoor security risks. Beijing then warned domestic companies to avoid the American chips. State-affiliated corporations and agencies received strong guidance to use domestic alternatives.

The H20 was designed to fall below export restriction thresholds. Nvidia's compromise offering, a chip optimized for the Chinese market within regulatory constraints. Beijing rejected it anyway.

Craig Singleton at the Foundation for Defense of Democracies captured the dynamic. "Chinese firms want H200s, but the Chinese state is driven by paranoia and pride," he told Reuters. "Paranoia about backdoors and dependence on U.S. chips, and pride in pushing domestic alternatives." The kicker: "Washington may approve the chips, but Beijing still has to let them in."

How much more capable is the H200? The Institute for Progress estimates it delivers nearly six times the performance of the H20. Georgetown University's Center for Security and Emerging Technology calculates it approaches ten times the previous export limit. That gap makes the chip more attractive to Chinese AI labs. It also makes Beijing's strategic concerns more acute.

The restrictions that began under Biden in 2022 lit a fire under China's domestic chip industry. Huawei has been grinding away at better AI processors. Beijing has poured money into local alternatives. The goal isn't better access to American silicon. The goal is not needing it at all.

Some analysts interpreted Beijing's H20 rejection as a negotiating tactic. Push back on the weaker chip to secure approval for something better. The H200 approval tests that theory. If China's objections were strategic theater, the better chip should find buyers. If the objections reflect genuine policy commitment to technological independence, the approval changes nothing.

Chris McGuire, who served on the National Security Council under Biden, believes Chinese firms will buy. "It would be self-defeating not to, given the H200 is better than every chip the Chinese can make," he told Reuters. But McGuire also called the export approval "very bad" for American interests. The contradiction isn't that both observations coexist. It's that the United States is selling its sword to an adversary who might be too proud to hold it. Washington gains nothing if Beijing refuses delivery. Washington loses if Beijing accepts.

Huang's Washington Strategy

The H200 approval represents Jensen Huang's most significant policy win since Nvidia became the world's most valuable company. It also demonstrates how deep he has cultivated relationships in the Trump administration.

Huang met with Trump at the White House last week to discuss export controls. Neither side disclosed details. Afterward, Trump praised the CEO. "He's done an amazing job," the president said. The approval followed within days.

The relationship dates to at least November 2024. Huang has built allies across the administration, including AI czar David Sacks and Commerce Secretary Howard Lutnick, both of whom backed H200 exports as a reasonable compromise. Secretary of State Marco Rubio and others torpedoed an earlier push for Blackwell exports, but the H200 found sufficient support.

Congressional critics have framed Nvidia's lobbying in less flattering terms. Senator John Kennedy of Louisiana refused to attend a recent meeting with Huang on Capitol Hill. "I don't consider him to be an objective, credible source about whether we should be selling chips to China," Kennedy said. He added that the CEO is "only interested in financial gain."

Senator Elizabeth Warren called the approval a result of "backroom meetings" and company donations. "Nvidia CEO Jensen Huang got his wish to sell the most powerful AI chip we've ever sold to China," she said. "This risks turbocharging China's bid for technological and military dominance."

Bipartisan opposition has proven ineffective. The SAFE Chips Act, introduced by Republican Pete Ricketts and Democrat Chris Coons, would codify existing export restrictions. The GAIN AI Act would have required American customers to receive priority access. Neither provision made it into must-pass defense legislation. Nvidia's lobbying operation outmaneuvered congressional skeptics.

Here's an awkward detail. Hours before Trump's announcement, the Justice Department revealed it had broken up a China-linked smuggling operation. The ring moved at least $160 million in controlled Nvidia H100 and H200 chips between late 2024 and early 2025. One arm of the government was prosecuting illegal chip exports. The other arm was legalizing them.

The Huawei Gambit

Administration officials defending the H200 approval have coalesced around a specific argument. Restricting American chip sales doesn't prevent Chinese AI development. It just shifts market share to Huawei.

The argument is simple: Chinese companies need AI chips. If Nvidia can't sell them, Huawei will. Huawei's chips are inferior, but they're improving. Every sale Nvidia loses accelerates Huawei's progress by providing revenue for research and development. American restrictions therefore accomplish the opposite of their intended effect.

This has become Nvidia's primary talking point. Huang has made it in Washington, to lawmakers, to administration officials, to anyone who will listen. The administration's internal deliberations framed the H200 approval as a compromise between Blackwell exports (which Trump declined to allow) and complete restriction (which officials believe would boost Huawei).

The argument contains a grain of truth wrapped in salesmanship.

Huawei is developing AI chips. Chinese demand exists regardless of American supply. But several assumptions deserve scrutiny.

First, Huawei's current offerings remain behind Nvidia's. Bloomberg reported that the H200 is "on paper at least a generation ahead of anything offered by Chinese designers from Huawei to Cambricon Technologies Corp. and Moore Threads Technology Co." The gap matters. Chinese AI labs using Nvidia chips can match American capability levels. Labs using domestic alternatives cannot.

Second, the argument assumes fungibility that may not exist. Huawei faces its own supply constraints. Taiwan Semiconductor Manufacturing Company won't fabricate its advanced designs. Chinese foundries lack equivalent capability. Even if demand shifted to domestic suppliers, those suppliers couldn't fill orders at scale.

Third, the logic proves too much. By this reasoning, any export restriction on any technology benefits foreign competitors. The argument would justify selling Blackwell, or Rubin, or whatever comes next. At some point, national security concerns must override commercial considerations. The question is where to draw that line, not whether lines exist.

The Institute for Progress report released Sunday found that H200 exports would allow Chinese AI labs to build supercomputers achieving performance similar to top American systems, "albeit at higher costs." Similar performance. The capability gap that export restrictions were designed to maintain would narrow.

Aaron Bartnick, a former White House technology and security official now at Columbia University, called the decision "shortsighted." The move will "advance China's chip capabilities," he said, and "it doesn't seem like the U.S. got much of consequence in return."

What the U.S. got was a 25% cut of sales that may never occur, collected through a mechanism that may not survive legal review, from a customer that may refuse to buy.

That's the deal.

Why This Matters

• For Nvidia investors, the approval creates headline value without guaranteed revenue. Beijing's response to actual chip availability will determine whether this translates to material financial impact. Watch for Chinese regulatory signals in coming weeks.
• For U.S. semiconductor policy, the H200 decision establishes a precedent that commercial lobbying can override bipartisan congressional opposition on export controls. Future restrictions face a higher political bar.
• For Chinese AI development, the question becomes whether Beijing's self-sufficiency goals or its labs' capability needs win the internal debate. Companies want the chips. The state has repeatedly signaled it doesn't want dependence on American technology. Something has to give.