The Tollbooth State

On August 10, 2025, the Trump regime crossed a threshold that history will not forget. Call this the Tollbooth State—or, just as aptly, the Tributary State—a merger of state and corporate power where every gate is priced, every river crossing exacts a cut, and every road runs through the executive’s hand.

Per Financial Times reporting, echoed by Axios and Reuters, Nvidia and AMD agreed to give the U.S. government 15% of their China AI-chip revenues—Nvidia’s H20 and AMD’s MI308—in exchange for Commerce Department export licenses. The New York Times, citing three people familiar with the arrangement, reported that the deal was personally struck between Nvidia CEO Jensen Huang and President Trump during a White House meeting. Two days later, Commerce began issuing the previously withheld licenses.

What is novel here is not that licenses can be granted or withheld; it is the price. BIS has historically charged no fee to file an application, and any authorized user fee must be tied to cost recovery—not to a flat percentage of private revenue. In National Cable Television Ass’n v. United States (1974) and FPC v. New England Power (1974), the Supreme Court made clear: such a condition reads as a tax, and Congress must speak clearly if it means to delegate taxing authority. Skinner v. Mid-America Pipeline (1990) confirms that point. Here, no statute grants BIS the power to demand a revenue share, and the Miscellaneous Receipts Act (31 U.S.C. § 3302) requires funds “received for the Government” to flow into Treasury absent explicit retention authority.

If this footing is missing, the entire scheme is ultra vires—no matter how novel the geopolitical gloss.

Export controls once carried the weight of principle. They existed to guard strategic advantage, not to auction it. To keep dangerous tools from hastening an adversary’s rise. That fiction has collapsed.

The Nvidia/AMD bargain did not ease because the danger waned; it eased because a price was named and met—with licenses flowing after a high-level reversal.

Mussolini’s corporazioni left industries nominally private, but every artery pulsed under state consent and its cut. Hitler’s ministries warped quotas and licenses until markets bent wholly to the Reich. The United States has its own, smaller-scale precedents—railroad land grants conditioned on political loyalty in the 19th century, and telecom spectrum allocations that became political patronage in the 1930s. In each, private ownership survived on paper, but operation bowed to executive will.

Here, the mask is similar: a “free market” whose hinges are owned by the executive. As The New York Times noted, the agreement effectively makes the federal government a partner in Nvidia’s China business — an arrangement with few precedents in U.S. export policy.

That separation—thin enough to see through, thick enough to deny—is precisely what distinguishes fascism from other authoritarian systems. In this light, the Nvidia/AMD revenue-share scheme may be the Trump regime’s most explicitly fascist act to date: a merger of state power and capitalist enterprise under a veneer of market legitimacy. Their attempted capture of Harvard’s governance offered a glimpse of the endgame: the selective integration of institutions into a state–corporate bloc that controls resources, knowledge, and legitimacy itself.

For Nvidia and AMD, the deal buys predictability—a guaranteed corridor into China—purchased at the cost of principle, underwritten by the future vulnerability of every industry that copies the move.

Once sold abroad, the principle is impossible to keep from being sold at home.

█ When export control is priced, the question is no longer whether something should be sold—only what tribute must be paid before the gate swings open. Corporations will meet the price; local economies will pay the remainder.

The immediate damage is the corrosion of trust in American export controls. Allies will know they can be outbid; adversaries will know they can buy their way through. Every future invocation of “national security” will register as an opening bid, not a line in stone.

For corporations, the toll is a manageable cost—spread to buyers or absorbed to protect a market. For the nation, it is the surrender of deterrence. Beijing no longer needs espionage to narrow the AI gap; it can purchase the keys.

Coalition risk follows fast: plurilateral export regimes and allied carve-outs fray when Washington signals that cash outranks control, undermining the very discipline those regimes require. The same fracture appeared in 1981 when grain-export restrictions on the USSR crumbled after a few states secured exemptions for domestic political reasons.

Once credibility collapses abroad, it invites extraction at home.

Legal faultline (why this “fee” may fail):
Agencies may levy user fees only when Congress authorizes them—and those fees must be tied to cost or value, not naked revenue skims. The Supreme Court drew this line in National Cable and FPC v. New England Power. Congress can delegate revenue-raising authority (Skinner), but absent such a delegation to Commerce/BIS for export-license revenue sharing, the condition is ultra vires. The Miscellaneous Receipts Act closes the loop: funds without retention authority must go to Treasury.

█ If this “fee” lacks statutory footing, it is not policy—it is a tax by license, and a ripe target for litigation.

What begins at the water’s edge will wash inland. If Washington can skim from corporations for the right to export, it can skim from states for the right to operate vital industries at home.

The method is identical: a license here, a permit there. A critical grid upgrade “delayed” until the “terms” are met. A pipeline expansion allowed only if a share of revenue—or loyalty—flows upward. In 1941, the War Production Board used this exact model to compel material allocation compliance from domestic factories; the “war need” rationale then, the “security license” rationale now.

Corporations have already swallowed the principle. Why resist its domestic clone? In a merged state-corporate economy, the license becomes a moat—keeping out smaller challengers, local innovators, and independent state-backed ventures.

█ Once tribute is normalized, a tollbooth—or a customs wall—can rise anywhere: at a port, a power line, a lab. Those who have already paid for one gate will lobby to keep others locked.

AI chips are only the trunk. Branches will sprout: semiconductor tools, agricultural hardware, biotech IP, aerospace, satellites, energy grids, rare earths. The licensing machinery already exists; the precedent now blesses its monetization.

In 2023, BIS processed over 48,000 export license applications. If even a fraction had been subject to a percentage-of-revenue toll, the sums would rival the budgets of entire agencies.

This is how medieval river tolls turned the Rhine into a gauntlet of petty lords, each extracting his cut. Trade slowed; towns withered. Only the Hanseatic League—cities bound in mutual refusal—broke the chain. Our states will need such a league, or watch the tollbooth metastasize into every artery of commerce.

█ When every road to market runs through a paid gate, the market ceases to be a commons. It becomes a corridor of tribute.

Beneath the cascade is the deeper rot. Once licensing is a tool for revenue and obedience, innovation is rewarded for compliance, not ingenuity. Contracts flow to the pliant, not the pioneering.

Blue cities and states out of political favor will see industries throttled, research bled, futures mortgaged to a gatekeeper they did not choose; individuals and institutions with a power and an inclination to subject the regime to friction will likewise be targeted. The corporations, welded to the licensing apparatus, will profit twice: securing their position and locking out all challengers beyond the regime’s reach.

█ The tollbooth does not just collect. It decides who may travel—and, over time, who may exist in the marketplace at all.

What it prices today, it will prohibit tomorrow.

States cannot dismantle the tollbooth once it becomes the federal business model. They must route around it.

Lock the procurement firewall: bar contracts with any vendor that accepts federal revenue-share conditions tied to export licensing. Under the market-participant doctrine, such state purchasing conditions survive Dormant Commerce Clause challenge. Vendors who comply with tribute schemes lose the right to bid; breach triggers immediate removal from the rolls.

Secure the intellectual commons: place university and lab IP in state-chartered trusts; forbid licensing contingent on federal tribute or political tests. Precedent: California’s stem cell IP policy insulating research from federal conditions.

Open new corridors: build state-to-province research partnerships, written with federal-law savings clauses to avoid preemption, to dilute BIS chokepoints. The 2014 California–Quebec climate compact offers the model.

Entrench an anti-tribute amendment: a constitutional or organic-law clause prohibiting compliance with license schemes conditioned on financial or political kickbacks, and requiring immediate legislative review of any federal license that redirects in-state revenues.

Align the litigation front: coordinate AGs to challenge revenue-share licensing as ultra vires under National Cable, Skinner, and the Miscellaneous Receipts Act; demand statutory authority and the agency’s OMB A-25 analysis.

Above all—build the league. Form an interstate Anti-Tollbooth Compact1 to mirror procurement bans, share IP-trust models, pool litigation resources, and synchronize discovery demands on license terms. Publish quarterly reports naming every federal licensing condition that imposes monetary or political tribute.

█ The Hanseatic cities knew no lone merchant could defy the toll lords. Build the league—or live by the gate.