Trump had been previewing the change before the executive order. At the United Nations General Assembly on September 22, he argued that “artificial intelligence” made the technology sound fake and said the United States would instead refer to it as “super intelligence.” He had previously floated alternatives including “extreme intelligence,” “superior intelligence” and “supreme intelligence,” before settling on the shorter and more marketable “super.”

A week later, explaining the decision again at the White House, Trump said the problem was specifically the word “artificial.” In his framing, today’s systems are no longer merely imitating intelligence. They are increasingly producing capabilities that deserve to be described as something more powerful.

The executive order formalizes that idea. It argues that modern systems have moved beyond what the original phrase “artificial intelligence” was meant to capture, increasingly amplifying human creativity and opening capabilities across science, medicine and other fields. The government’s terminology, the order says, should reflect that transformation. Taken literally, this is a debate over semantics. Taken more seriously, it is an attempt to reposition an entire technological category.

The word artificial has always carried a slight tension. Artificial flowers are not flowers. Artificial flavors are not natural flavors. Artificial light is not sunlight. The word often describes something manufactured as a substitute for the real thing. For decades that was an accurate description of the ambition behind artificial intelligence: machines attempting to reproduce behaviors associated with human intelligence.

But AI is no longer perceived only as imitation. Large models write software, generate video, interpret medical data, search scientific spaces and increasingly operate software autonomously. Whether these systems should be considered genuinely “intelligent” is still a philosophical and scientific question, but their economic significance no longer depends on resolving it. Trump’s rebranding takes advantage of that ambiguity. “Artificial intelligence” describes where the field came from. “Super intelligence” describes where its advocates want people to believe it is going.

There is another reason the timing matters. “AI” is currently one of the most valuable terms in global business, but it has also accumulated a remarkable amount of baggage.

Over only a few years, AI has become associated with job displacement, copyright disputes, deepfakes, misinformation, data-center electricity consumption, surveillance, algorithmic bias, autonomous weapons and fears of machines becoming difficult to control. The same two letters that attract investment can also trigger public anxiety.

Trump’s administration has taken a distinctly optimistic approach to that debate, emphasizing technological leadership, economic expansion and competition with China while resisting calls to broadly slow development. At the September 29 White House event, leading technology companies signed a voluntary commitment around frontier-system responsibilities, while Trump continued to frame rapid development as strategically important to the United States.

Seen in that context, replacing AI with SI does more than remove an unpopular adjective. It changes the emotional direction of the term.

Artificial Intelligence asks whether machines can reproduce intelligence.

Super Intelligence suggests that intelligence itself is becoming more capable.

The first is descriptive. The second is aspirational. And that difference may be exactly the point.

The rebrand becomes considerably more complicated once it leaves politics and enters computer science. “Superintelligence” is not a blank term.

In AI research and futurist literature, it has traditionally described a hypothetical stage at which machine intelligence substantially exceeds human cognitive performance across a broad range of domains. Discussions of advanced AI have often used a rough conceptual progression from narrow AI to artificial general intelligence, or AGI, and eventually to artificial superintelligence, often abbreviated ASI.

These categories are not universally defined, and researchers disagree over where one ends and another begins. But “superintelligence” has generally referred to something considerably more ambitious than the AI systems currently deployed.

The Trump administration is doing something very different. Under the September 29 executive order, “SI” initially refers to essentially the same class of technologies already covered by the statutory definition of AI. That creates an unusual semantic collision.

A chatbot becomes SI.
An image generator becomes SI.
An enterprise recommendation engine could, in government terminology, become SI.

And someday, if a system actually reaches what researchers previously called superintelligence, the industry may need another word to distinguish it.

This is one reason AI is unlikely to disappear as a technical term. It is embedded in decades of research papers, academic departments, international standards, legislation, software documentation and corporate infrastructure. The European Union has an AI Act, companies operate AI labs and engineers build AI systems. A U.S. executive order can change the vocabulary of the executive branch. It cannot instantly rewrite the vocabulary of computer science. But that does not mean SI has to replace AI to succeed.

That process may already be beginning. On October 4, Elon Musk said SpaceX would rename its AI operation from SpaceXAI to SpaceXSI, explicitly following the emerging “super intelligence” terminology. The Trump administration has also launched a federal Super Intelligence Force, reinforcing the new terminology inside government institutions rather than treating it as a one-off rhetorical flourish. This is where the experiment becomes more interesting.

Political terminology rarely becomes technological terminology simply because a government orders it. But language can spread when companies discover commercial value in it. “Super intelligence” sounds newer than AI. It sounds more advanced than generative AI. It implies a technological frontier without requiring a company to make the far more specific claim that it has achieved AGI. That leaves room for a new marketing vocabulary:

SI infrastructure.
SI agents.
SI factories.
SI computing.
SI platforms.

Nvidia CEO Jensen Huang has already used the phrase “superintelligence factories” in the broader discussion surrounding the terminology. If more companies follow Musk, SI may not become the formal scientific replacement for AI. It could instead become the commercial name for the next layer above ordinary AI — less a technical classification than a signal that a product belongs to the frontier.

Technology history is full of terms that became powerful precisely because their definitions were flexible. “Cloud” originally described a particular computing architecture before becoming an enormous commercial category. “Web3” accumulated companies, investment and products long before there was agreement on precisely what belonged inside it. “Metaverse” briefly became a corporate strategy, an investment thesis and a product label at the same time. SI could follow a similar path.

The strangest part of the story arrived almost immediately after Trump began promoting the new abbreviation.

The country-code top-level domain for Slovenia is .si. And Melania Trump was born in Slovenia. That coincidence was almost too perfect for the internet to ignore.

After Trump promoted “super intelligence” at the United Nations, demand for Slovenian .si domains surged. Registry data cited by TechCrunch showed a 2,199% increase in .si purchases in September, while other reporting documented thousands of registrations arriving within days of Trump’s remarks. Netcraft found that .si registrations briefly outpaced .ai registrations in the 25 hours following Trump’s UN announcement, with evidence that a meaningful portion of the buying was speculative. More than half of some domains purchased around the announcement were subsequently listed for resale.

The parallel with .ai is obvious. The .ai domain is not technically an AI domain. It is the country-code domain of Anguilla, a small British Overseas Territory in the Caribbean. The artificial-intelligence boom transformed that otherwise geographically specific domain into one of technology’s most recognizable digital assets. Now Slovenia may be experiencing a miniature version of the same phenomenon.

A company called something.ai immediately signals artificial intelligence, regardless of whether anyone knows where Anguilla is. If SI develops a durable association with super intelligence, .si could acquire the same semantic shortcut. And because Melania Trump is Slovenian, an inevitable theory appeared almost instantly: was the entire SI rebrand somehow designed around Slovenia?

There is currently no credible evidence supporting that explanation. Trump has publicly explained his reasoning repeatedly, and his explanation has consistently focused on his dislike of the word “artificial.” The White House executive order makes the same argument. None of the available evidence indicates that Melania Trump or Slovenia’s country domain played a role in selecting the term.

The Slovenia connection is therefore better understood as an extraordinary coincidence — and an unexpectedly profitable one for domain speculators — rather than an established explanation for the policy. But the fact that people immediately noticed it says something important about how rapidly a new technology label can acquire meaning. Within days, SI went from an unusual presidential phrase to an abbreviation associated with government policy, technology branding, SpaceX and a suddenly fashionable European domain extension.

The easiest interpretation of Trump’s executive order is that the United States is simply renaming AI. The more interesting interpretation is that we may be watching the beginning of two overlapping vocabularies. 

The first will remain technical and international. Researchers will continue publishing AI papers. Universities will continue teaching artificial intelligence. Governments outside the United States will continue writing AI legislation. Developers will continue calling APIs from AI companies.

The second vocabulary may become increasingly commercial and political. In that world, AI describes the underlying technology, while SI describes the frontier — the systems, infrastructure and ambitions meant to represent what comes next. That distinction could ultimately be more influential than a complete rename.

The technology industry constantly needs new language because successful technologies become ordinary. “AI-powered” once sounded advanced; today it appears on everything from enterprise software to toothbrushes. As AI becomes ubiquitous, companies building the most expensive and capable systems need another way to communicate that they are working on something beyond the commodity layer. “Super intelligence” provides exactly that. Whether it becomes scientifically meaningful is almost secondary.

The most revealing part of Trump’s SI experiment may therefore have little to do with whether people stop saying AI. Technological eras are not defined only by inventions. They are also defined by the language societies build around those inventions. “Personal computer,” “internet,” “smartphone,” “cloud,” “social media,” “metaverse” and “artificial intelligence” did more than describe technologies. Each term created a category large enough for companies, investment, policy and culture to organize around it. For seventy years, artificial intelligence has been one of those categories.

Trump is now attempting something unusually ambitious: renaming the category while the technological transformation is still accelerating. The effort may fail. AI is deeply embedded in global language and scientific history, and “superintelligence” already has a technical meaning that does not neatly correspond with today’s models. But SI does not need to eliminate AI to matter.

If governments repeat it, companies adopt it, investors recognize it and consumers begin associating it with the frontier of computing, then it becomes a brand before it ever becomes a scientific definition. That may ultimately be the more important story. Artificial intelligence is not suddenly becoming superintelligence.

Super intelligence is becoming a brand.