Microsoft’s AI chief says superintelligence is near, but won’t take your job
Mustafa Suleyman says Microsoft can now pursue its own frontier AI while keeping its OpenAI ties. He argues superintelligence is close, but says the company’s goal is useful products, not replacing people.
Intelligence analysis by GPT-5.4 Mini
In a Decoder interview, Microsoft AI CEO Mustafa Suleyman says a new OpenAI contract freed Microsoft to build its own frontier models and a Superintelligence team. He also defends the partnership’s evolution, criticizes AI being described as conscious, and says the real test is whether consumer products are good enough to overcome public skepticism.
Microsoft’s AI boss says the company is building very smart computer brains of its own, even while still working with OpenAI. He thinks these systems can become powerful fast, but says they should be treated like tools, not like living beings or job-stealing robots.
Analysis
Microsoft’s new AI posture
Mustafa Suleyman says Microsoft has spent the last 15 to 18 months reshaping its relationship with OpenAI. According to him, a new contract signed in October of last year both strengthened the partnership and gave Microsoft room to pursue superintelligence independently while still buying and licensing OpenAI models.
That change matters because Suleyman says it let him focus on building a Superintelligence team, assembling large enough compute clusters for frontier training, and hiring around that mission. He ties that shift to the new models Microsoft announced at Build, describing it as a major reorientation inside the company.
OpenAI, consumer products, and the frontier
Suleyman frames the OpenAI relationship as an unusually successful one that has simply entered a new phase. He says OpenAI’s ambitions naturally broadened as it gained traction and revenue: it moved into consumer products, its own data centers, chips, and direct-to-market offerings like ChatGPT Enterprise. That expansion, in his telling, explains why Microsoft now wants more control over its own model development.
The interview also touches on public skepticism. Suleyman says Microsoft is thinking hard about negative polling and political pushback around AI, and whether consumer products can become strong enough to overcome it.
His warning on AI language
Suleyman also pushes back on language that makes systems sound conscious or alive. He calls that framing dangerous, which signals Microsoft’s effort to keep its AI messaging centered on tools and utility rather than personhood or mystique.
Key points
- Suleyman says a new contract with OpenAI gave Microsoft room to pursue superintelligence independently.
- He says he has been assembling a Superintelligence team and building the compute needed to train frontier models.
- He describes the Microsoft-OpenAI relationship as highly successful, but now entering a new stage.
- He says OpenAI’s move into consumer products, chips, and data centers helped drive the broader shift.
- He argues it is dangerous to describe AI as conscious or alive.
If Microsoft’s new frontier-model push works, it could produce stronger AI products across text, images, audio, and video. The company also seems to believe better consumer tools could reduce public skepticism by making the benefits easier to see.
The split between Microsoft and OpenAI could deepen competition and make the partnership harder to manage over time. Suleyman’s remarks also show the company is still facing political pushback and unease about AI, which could slow adoption if consumers remain unconvinced.
