For decades, we have called it artificial intelligence: AI. Now Donald Trump wants the U.S. government to call it Super Intelligence, or SI. His September 29, 2026 executive order directs federal executive departments and agencies to adopt the new terminology.
Government websites are already making the switch
The change is already visible at the Department of Energy. Its October 9 science fact sheet describes “Super Intelligence-enabled” research and uses SI when discussing scientific datasets, robotics, and laboratory automation. It also announces 18 additional Genesis Mission awards, bringing the first-year portfolio to 297 projects. The new name now accompanies concrete research programs and funding announcements.
Other federal bodies are following. NIST, part of the Department of Commerce, explicitly says it is updating its communications to comply with the order. NASA’s website carries a similar notice and navigation labels including “What is SI?” and “SI Ethics.” Neither presents the terminology change as evidence of a newly achieved intelligence threshold.
For developers reading these pages, that distinction matters. An official website saying “Super Intelligence” can sound like an announcement that computers have surpassed human intelligence. The documented change is a naming policy, with implications for how technology is discussed, marketed, and understood.
Why Trump wants a different name
Trump’s campaign against the name predates the order. In July 2025, he complained that “artificial” was an unsuitable description of technology he regarded as genius. His September 19, 2026 X post proposed alternatives including Superior, Extreme, and Supreme Intelligence. PolitiFact’s reporting documents these posts and the earlier remarks. The objection was about what the name communicates: “artificial” sounded dismissive of the technology’s value.
The administration’s formal explanation adds a national competitiveness argument. Its September 29 fact sheet presents SI as language that better expresses technological progress, American leadership, and potential gains in science and the economy. My reading is that the rebrand serves two purposes: replacing an association with imitation and giving technological expansion a more ambitious public identity. Those are communications goals; neither establishes a new capability.
What the order actually changes
Executive Order 14434, signed September 29, directs executive departments and agencies to replace AI with SI in communications, websites, reports, and other non-statutory documents, where legally permitted. It does not impose that vocabulary on every developer or private company. Previously issued contracts, grants, regulations, and historical documents do not have to be rewritten.
The most revealing detail is Section 3: initially, SI covers the systems already encompassed by the existing statutory definition of artificial intelligence. The order also requests proposed legislative language for a future definition within 60 days. The immediate change therefore renames an existing category without demonstrating a corresponding jump in performance.
Superintelligence already means something
Superintelligence already had a technical meaning long before this administration. In his 1998 paper, Nick Bostrom described an intellect substantially exceeding the best human minds across practically every field. The definition concerns breadth and degree of capability, while leaving implementation and consciousness open.
A useful distinction is that AI names a broad field, AGI concerns general capabilities across tasks, and artificial superintelligence describes capabilities beyond human levels. Definitions remain contested, but these terms are not interchangeable. Google DeepMind researchers’ Levels of AGI paper separates generality from performance: exceptional achievement in one narrow domain does not automatically establish general superintelligence. That distinction survives any government rebrand.
Big Tech’s response is mixed
Big Tech’s response also needs chronology. Some executives have embraced Trump’s preferred wording. Elon Musk promised to change SpaceXAI to SpaceXSI, according to reporting on his X response. A promised rename should not be confused with a completed change across products, documentation, and corporate identities.
Nvidia provides a concrete example of adoption in company communications. Its October 8 announcement describes commitments valued at $1 billion over five years to advance American science, using “super intelligence” for research involving quantum computing, healthcare, and energy security. This shows the phrase moving beyond presidential speeches into an investment announcement, although it does not establish that the wording caused the investment.
Other companies had already adopted superintelligence as an aspiration. Meta published “Personal Superintelligence for Everyone” in July 2025. Microsoft announced its Humanist Superintelligence approach in November 2025, emphasizing practical capabilities under human control. Both precede the September 2026 order. Treating every corporate use as compliance with Trump would erase that history. Meanwhile, PolitiFact found limited change in public-facing terminology at OpenAI, Google DeepMind, and Anthropic. There is no basis here for declaring an industry-wide replacement.
The naming announcement also accompanied a separate governance agreement. Associated Press reported that leaders from Google, Meta, Anthropic, OpenAI, Nvidia, and Musk’s company signed a voluntary accord involving internal and external reviews. This matters because adopting a vocabulary and accepting oversight are different decisions. An executive’s presence at the event cannot establish agreement with every technical claim attached to the new name.
After Super, will we get Mega and Hyper?
A LinkedIn comment shared during preparation of this article offers an interesting branding interpretation: AI sounds artificial and open-ended, while SI creates room for future upgrades such as Mega Intelligence or Hyper Intelligence.
That observation is useful as marketing analysis, but its technical premise needs correction. “Artificial” describes an engineered origin; it does not rule out human-like behavior. Nor does AI formally mean a technology with “no start or end.” The field can contain successive generations, architectures, and capability levels without changing its umbrella name.
Super, Mega, and Hyper could become product tiers, much like other escalating commercial labels. However, without agreed tests, those tiers would communicate positioning rather than measurable intelligence. A vendor could upgrade the name while leaving the underlying model unchanged. The useful question is what improved: task completion, reliability, generalization, operating cost, or the amount of supervision required.
What developers and users could face
Several consequences are plausible. First, customers may interpret “super” as a guarantee of competence. Imagine a coding assistant promoted as SI: a team might grant it broader repository access or reduce review because the label suggests superior judgment. That is a hypothetical risk, not evidence that this rebrand has already changed behavior. The appropriate safeguard is to base permissions on demonstrated performance and the consequences of failure.
Second, mixed terminology could create practical friction. Researchers, procurement teams, and documentation maintainers may need to search for both AI and SI, preserve older references, and explain which meaning applies. Government-facing documentation could define SI as the agency’s terminology for AI, then describe the actual system precisely.
Search and retrieval systems may need to recognize both terms while preserving the source’s wording and date. Otherwise, searches for AI could miss newer government material, while searches for superintelligence could return research with much narrower ambitions.
Third, policy debate could become harder to follow. When the same phrase denotes ordinary deployed tools and hypothetical broadly superhuman systems, readers can mistake discussions of one for claims about the other. Future proposals should be evaluated by their definitions, covered systems, and obligations. A more impressive noun does not resolve those boundaries.
There is a potential benefit: ambitious language can attract attention to useful scientific work and make its goals easier to communicate. But that benefit depends on keeping aspiration distinguishable from achievement. Calling a research program SI should invite a clearer explanation of its aims, evidence, and limitations.
For developers, the practical response is to preserve that clarity. State the model, task, evaluation conditions, observed failures, and human oversight. If a customer requests SI terminology, define it explicitly. When someone announces the next upgrade to Mega or Hyper Intelligence, ask for the evaluation that changed—and check whether it measures the work your users actually need done.
