“We negotiated a little deal,” President Donald Trump told reporters on August 11, about the developing situation with leading chip makers Nvidia and AMD continuing to do business in China. He explained that he originally wanted a 20% cut of Nvidia’s sales in exchange for the company obtaining export licenses to sell H20 chip to China, but he was persuaded to settle at 15%. The H20 chip is “obsolete,” Trump added … “he’s selling a essentially old chip.”
The chips do appear to be quite significant to China, considering that the Cyberspace Administration of China held discussions with Nvidia over security concerns that the H20 chips may be tracked and turned off remotely, according to a disclosure on its website. The deal, which lifted an export ban on Nvidia’s H20 AI chips and AMD’s MI308, and followed heated negotiations, was widely described as unusual and also still theoretical at this point, with the legal details still being ironed out by the Department of Commerce. Legal experts have questioned whether the eventual deal would constitute an unconstitutional export tax, as the U.S. Constitution prohibits duties on exports. This has come to be known as the “export clause” of the constitution.
Indeed, it’s hard to find much precedent for it anywhere in the history of the U.S. government’s dealings with the corporate sector. Erik Jensen, a law professor at Case Western Reserve University who has studied the history of the export clause, told Fortune he was not aware of anything like this in history. In the 1990s, he added, the Supreme Court struck down two attempted taxes on export clause grounds (cases known as IBM and U.S. Shoe). Jensen said tax practitioners were surprised that the court took up the cases: “if only because most pay no attention to constitutional limitations, and the Court hadn’t heard any export clause cases in about 70 years.” The takeaway was clear, Jensen said: “The export clause matters.”
Columbia University professor Eric Talley agreed with Jensen, telling Fortune that while the federal government has previously applied subsidies to exports, he’s not aware of other historical cases imposing taxes on selected exporters. Talley also cited the export clause as the usual grounds for finding such arrangements unconstitutional.
Rather than downplaying the uniqueness of the arrangement, Treasury Secretary Scott Bessent has been leaning into it. In a Bloomberg television interview, he said: “I think you know, right now, this is unique. But now that we have the model and the beta test, why not expand it? I think we could see it in other industries over time.”
Bessent and the White House insist there are “no national security concerns,” since only less-advanced chips are being sold to China. Instead, officials have touted the deal as a creative solution to balance trade, technology, and national policy.
The arrangement has drawn sharp reaction from business leaders, legal experts, and trade analysts. Julia Powles, director of UCLA’s Institute for Technology, Law & Policy, told the Los Angeles Times: “It ties the fate of this chip manufacturer in a very particular way to this administration, which is quite rare.” Experts warned that if replicated, this template could pressure other firms—not just tech giants—into similar arrangements with the government. Already, several unprecedented arrangements have been struck between the Trump administration and the corporate sector, ranging from the “golden share” in U.S. Steel negotiated as part of its takeover by Japan’s Nippon Steel to the federal government reportedly discussing buying a stake in chipmaker Intel.
Nvidia and AMD have declined to comment on specifics. When contacted by Fortune for comment, Nvidia reiterated its statement that it follows rules the U.S. government sets for its participation in worldwide markets. “While we haven’t shipped H20 to China for months, we hope export control rules will let America compete in China and worldwide. America cannot repeat 5G and lose telecommunication leadership. America’s AI tech stack can be the world’s standard if we race.”
The White House declined to comment about the potential deal. AMD did not respond to a request for comment.
While Washington has often intervened in business—especially in times of crisis—the mechanism and magnitude of the Nvidia/AMD deal are virtually unprecedented in recent history. The federal government appears to have never previously claimed a percentage of corporate revenue from export sales as a precondition for market access. Instead, previous actions took the form of temporary nationalization, regulatory control, subsidies, or bailouts—often during war or economic emergency. Examples of this include the seizure of coal mines (1946) and steel mills (1952) during labor strikes, as well as the 2008 financial crisis bailouts, where the government took equity stakes in large corporations including two of Detroit’s Big three and most of Wall Street’s key banks. During World War I, the War Industries Board regulated prices, production, and business conduct for the war effort.
Congress has previously created export incentives and tax-deferral strategies (such as the Domestic International Sales Corporation and Foreign Sales Corporation Acts), but these measures incentivized sales rather than directly diverting a fixed share of export revenue to the government. Legal scholars stress that such arrangements were subjected to global trade rules and later modified after international complaints.
The U.S. prohibition on export taxes dates back to the birth of the nation. Case Western’s Jensen has written that some delegates of the Constitutional Convention of 1787, such as New York’s Alexander Hamilton, were in favor of the government being able to tax revenue sources such as imports and exports, but the “staple states” in the southern U.S. were fiercely opposed, given their agricultural bent, especially the importance of cotton at that point.
Still, many other countries currently have export taxes on the books, though they are generally imposed across all exporters, rather than as one-off arrangements that remove barriers to a specific market. And many of the nations with export taxes are developing countries who tax agricultural or resource commodities. In several cases (Uganda, Malaya, Sudan, Nigeria, Haiti, Thailand), export taxes made up 10% to 40% of total government tax revenue in the 1960s and 1970s, according to an IMF staff paper.
Globally, most countries tax profits generated within their borders (“source-based corporate taxes”), but rarely as a direct percentage of export sales as a market access precondition. The standard model is taxation of locally earned profits, regardless of export destination; licensing fees and tariffs may be applied, but not usually as a fixed percent of export revenue as a pre-negotiated entry fee.
Although the Nvidia/AMD deal doesn’t take the usual form of a tax, Case Western’s Jensen added. “I don’t see what else it could be characterized as.” It’s clearly not a “user fee,” which he said is the usual triable issue of law in export clause cases. For instance, if goods or services are being provided by the government in exchange for the charge, such as docking fees at a governmentally operated port, then that charge isn’t a tax or duty and the Export Clause is irrelevant. “I just don’t see how the charges that will be levied in the chip cases could possibly be characterized in that way.”
Players have been known to “game” the different legal treatments of subsidies and taxes, Columbia’s Talley added. He cited the example of a government imposing a uniform, across-the-board tax on all producers, but then providing a subsidy to sellers who sell to domestic markets. “The net effect would be the same as a tax on exports, but indirectly.” He was unaware of this happening in the U.S. but cited several international examples including Argentina, India, and even the EU.
One famous example of a canny international tax strategy was Apple’s domicile in Ireland, along with so many other multinationals keeping their international profits offshore in affiliates in order to avoid paying U.S. tax, which at the time applied to all worldwide income upon repatriation. Talley said much of this went away after the 2018 tax reforms, which moved the U.S. away from a worldwide corporate tax, with some exceptions.
If Trump’s chip export tax is an anomaly in the annals of U.S. international trade, the deal structure has some parallels in another corner of the business world: organized crime, where “protection rackets” have a long history. Businesses bound by such deals must pay a cut of their revenues to a criminal organization (or parallel government), effectively as the cost for being allowed to operate or to avoid harm.
The China chip export tax and the protection rackets extract revenue as a condition for market access, use the threat of exclusion or punishment for non-payment, and both may be justified as “protection” or “guaranteed access,” but are not freely negotiated by the business.
“It certainly has the smell of a governmental shakedown in certain respects,” Columbia’s Talley told Fortune, considering that the “underlying threat was an outright export ban, which makes a 15% surcharge seem palatable by comparison.”
Talley noted some nuances, such as the generally established broad statutory and constitutional support for national-security-based export bans on various goods and services sold to enumerated countries, which have been imposed with legal authority on China, North Korea, Iraq, Russia, Cuba, and others. “From an economic perspective, a ban on an exported good is tantamount to a tax of ‘infinity percent’ on the good,” Talley said, meaning it effectively shuts down the export market for that good. “Viewed in that light, a 15% levy is less (and not more) extreme than a ban.”
Still, there’s the matter, similar to Trump’s tariff regime, of making a legal challenge to an ostensibly blatantly illegal policy actually hold up in court. “A serious question with the chips tax,” Case Western’s Jensen told Fortune, “is who, if anyone, would have standing to challenge the tax?” In other words, it may be unconstitutional, but who’s actually going to compel the federal government to obey the constitution?
This story was originally featured on Fortune.com