The State, the Strait, and the Server Farm

Trump’s governing instinct, as it appears across this week’s dispatches, is less a doctrine than a procurement system for leverage. International order, copyright, local permitting and even geographic names are treated as things whose prior status is merely an opening bid. An Al Jazeera opinion essay calls this a turn from multilateral governance toward bilateral, transactional statecraft; that is the author’s interpretation, but it captures the administrative mood neatly.

The foreign-policy version uses tariffs, sanctions and alliance pressure under the banner of “peace through strength,” according to the essay. Its larger claim is that such conduct may encourage middle powers and regional blocs to build alternatives to Western-led institutions. That is forecast rather than settled fact—but institutions are, among other things, habits, and habits do not survive indefinitely when their largest participant starts treating every meeting as a ransom negotiation.

The domestic version is not gentler. The White House is pushing faster AI-infrastructure construction while the Justice Department urges a broad fair-use approach to training models on copyrighted work. In both cases, a diffuse public claim—local control, creator compensation, a judicially defined property right—encounters a national imperative with a flag pinned to it. The state is not abolishing the argument. It is deciding which party must have the argument after the cranes and models are already operational.

On Wednesday, Trump suggested renaming the Strait of Hormuz the Trump Strait, after signing an executive order the prior weekend deeming Lake Ontario “Lake America,” a change Apple and Google adopted for US map users, New York Magazine reported. The distinction between a domestic naming directive and a waterway partly bounded by Oman and Iran is, in conventional geography, fairly consequential.

The report also says Trump had previously suggested declaring Hormuz a US territory, without explaining how. No formal action to rename the strait or make it US territory is reported. Still, the episode is useful because it makes a governing habit unusually visible: symbolic assertion is deployed not as decoration but as a test of whether reality’s clerks—platforms, agencies, allies, perhaps eventually everyone else—will update the database.

Map apps are the perfect small empire for this purpose. They are private interfaces through which millions encounter public reality; change the label for one national audience and the sovereignty claim becomes a user experience. The strait, meanwhile, remains an actual shipping lane with actual territorial waters, not a resort wing awaiting a rebrand. The comedy is bureaucratic. The consequence is that the line between naming power and exercising power becomes easier to blur.

The Justice Department has filed a 20-page statement of interest in the consolidated copyright litigation involving OpenAI, The New York Times, other news outlets, authors and publishers. It argues that the court should reject the proposition that training large language models on copyrighted texts violates copyright law, framing the issue around US competitiveness and national security, Deadline reported. This is not a court ruling; the central fair-use question remains unresolved.

The government’s premise is admirably uncluttered by literary romanticism:

“Rules of law that make it significantly more difficult to develop a robust AI industry in the United States therefore threaten national security and give a competitive advantage to foreign adversaries who are not so encumbered.” — the DOJ filing

That argument converts a dispute over permission and compensation into a race against adversaries. It may be legally persuasive or not; a judge will decide that. Politically, however, it is an old move in new silicon packaging. Once an industry is described as strategic infrastructure, the costs imposed on particular people begin to look like regrettable but manageable externalities. The Times’ position, as reported, is that AI companies should compensate creators whose work underpins their products. The administration has chosen which side of that distributional question it sees as the national champion.

Judge Sidney H. Stein has ordered the plaintiff to explain by September 11 why the New York action should not be stayed pending summary-judgment motions in related multidistrict cases; defendants’ responses are due September 18. Procedure, bless it, is still asking for paperwork while grand strategy tries to vault the courthouse steps. Yet the direction of travel is plain enough: AI is being treated less like a consumer technology than like a strategic sector entitled to legal room, and perhaps to the cultural raw materials already produced by everyone else.

There is a material problem with calling AI infrastructure a national-security necessity: it has to be built somewhere, and “somewhere” usually has voters, aquifers and zoning rules. More than half of Americans used an AI chatbot in July, according to Morning Consult data cited by Business Insider, while a May Gallup poll found 70 percent opposed building AI data centers in their own communities. The familiar consumer-citizen split is doing brisk business.

Trump has warned resistant communities that they risk economic decline, even as the White House says he wants facilities only in communities that want them and requires centers to pay for their utilities. The administration has also pursued a permitting fast track, directed the Energy Department to identify federal land, and had the EPA propose removing notice-and-comment requirements for certain minor air-pollution permits that can include backup generators. The policy architecture offers local consent at the front door and a considerable amount of federal acceleration around the side.

Hood County, Texas, offers the inconvenient physical specimen. The county, where Trump won more than 80 percent of the 2024 vote, tightened disclosure requirements around proposed water and energy use and revoked a concept plan for an 862-acre, 100-megawatt project projected to use up to 20,000 gallons of water daily. One proposal envisioned 2,000 construction workers at peak and 200 permanent workers. These are not necessarily arguments against every data center; they are the arithmetic local politics insists upon when a national strategy arrives with a substation.

That arithmetic is now visibly bipartisan. Pennsylvania’s Josh Shapiro moved toward tougher transparency and local-approval requirements, while Texas’s Greg Abbott issued directives and a temporary development pause amid opposition. In Michigan, Democratic Senate nominee Abdul El-Sayed has likewise identified data centers as an issue Trump voters care about, in an interview that advances his broader critique of corporate power and emphasizes grassroots campaigning, The Nation reported. His claims are campaign arguments, not polling results, but the political opening is real: the AI boom has supplied both parties with a way to say they favor innovation while asking why innovation’s bills keep arriving at someone else’s house.

The deeper tension is not whether America wants AI. It plainly does. It is whether national power can keep defining itself as frictionless scale when scale requires land, water, electricity, copyrighted expression and public acquiescence. A renamed map may obey instantly. A county commission, a federal judge and an overloaded grid are less susceptible to the magic of the Sharpie.