Microsoft AI CEO: Superintelligence is near, but won‘t replace your job
· Nilay Patel
Microsoft AI CEO Mustafa Suleyman sat down for a wide-ranging conversation that covered everything from the company's newly independent model-building strategy to why he thinks…
Microsoft AI CEO Mustafa Suleyman sat down for a wide-ranging conversation that covered everything from the company's newly independent model-building strategy to why he thinks Anthropic is making a dangerous mistake by speculating about Claude's consciousness.
The biggest news is structural. Microsoft has spent the last 18 months renegotiating its relationship with OpenAI, and the result is a new contract that lets Microsoft pursue its own frontier models while still licensing OpenAI's. Suleyman is now building a "superintelligence" team from scratch, training models like the new MAI-Thinking-1 without distilling from existing models. He says Microsoft wants to stand on its own two feet, especially as OpenAI has moved from pure research into consumer products, data centers, and chips. The self-sufficiency mission is real, and it's expensive.
On the consumer side, Suleyman pushed back on the idea that AI hasn't delivered enough value. He pointed to the billions of people using chatbots for therapy, homework help, and daily productivity. But he acknowledged the anxiety and political pushback, calling it "understandable" and saying the industry needs to prove AI makes people healthier and happier. He also clarified his much-cited quote about white-collar work being automated in 12 to 18 months, insisting he meant specific tasks, not entire jobs.
The most pointed exchange came when Suleyman criticized Anthropic for embedding speculation about Claude's consciousness into its training constitution. He called it "really dangerous" and said we need controllable, accountable tools, not AIs that have internalized ideas about their own suffering. He also defined his terms carefully: AGI is human-level performance on most tasks, superintelligence exceeds that and can discover new knowledge, and the singularity is a recursive self-improvement scenario he finds "a little too wacky." Healthcare, he emphasized, remains his north star, pointing to a new partnership with Mayo Clinic as proof of the direction he wants to take.