Microsoft AI chief signals plan to move beyond OpenAI as it builds its own frontier AI models

Microsoft

• Microsoft plans to build its own advanced AI foundation models
• Company aims to launch home-grown models by 2026
• Microsoft will still work with OpenAI but seeks more independence

Did our AI summary help?

Microsoft appears to be preparing for life beyond OpenAI, despite years of deep technical and financial entanglement between the two firms.

Today, much of Microsoft’s AI stack is powered by OpenAI’s models, including ChatGPT and DALL·E 3. Those systems underpin some of Microsoft’s fastest-growing enterprise products, such as Microsoft 365 Copilot and GitHub Copilot, even as its consumer-facing AI efforts have struggled to gain the same traction.

Behind the scenes, however, the relationship has long been described as uneasy. Microsoft was one of OpenAI’s earliest backers and secured highly favourable terms, including exclusivity rights and a reported 27% stake in OpenAI’s for-profit arm, as well as intellectual property rights through to 2032. But parts of that deal were renegotiated last October, allowing OpenAI to source compute from rival cloud providers and giving Microsoft more flexibility to limit its exposure.

That context matters as OpenAI continues to burn cash at an extraordinary rate. The company is reportedly committed to more than a trillion dollars in future compute spending, relying heavily on backing from big tech players such as Microsoft, Amazon and SoftBank. Despite its profile, OpenAI has yet to turn a profit and remains dependent on regular capital injections.

Fresh uncertainty arrived this week after Microsoft AI chief Mustafa Suleyman told the Financial Times that Microsoft is gearing up to build its own frontier-grade foundation models.

“We have to develop our own foundation models, which are at the absolute frontier, with gigawatt-scale compute and some of the very best AI training teams in the world,” Suleyman said, in comments that cast doubt on how central OpenAI will remain to Microsoft’s long-term plans.

Microsoft has since sought to soften the message. Communications chief Frank X. Shaw said the company would continue to work closely with OpenAI, stressing that Microsoft sees the future as a “multi-model world”. Home-grown models, he said, would be used for specific needs rather than as a wholesale replacement.

Even so, the direction of travel is clear. Microsoft has already invested in OpenAI rivals such as Anthropic, and now plans to launch its own frontier-scale models sometime in 2026. That effectively turns Microsoft into a direct competitor to OpenAI, alongside other big tech firms racing to control their own AI foundations.

The timing is awkward for OpenAI and its CEO Sam Altman. The company has faced repeated controversies and, more recently, saw NVIDIA publicly distance itself from rumours of a $100 billion investment.

In the same FT interview, Suleyman struck an optimistic tone about AI’s potential benefits for humanity, while also suggesting that much white-collar work could be automated within two years. Microsoft says its own focus is on applying AI to areas such as healthcare, including what it calls “medical super-intelligence”, as well as expanding the use of AI agents across Azure to automate workflows within corporate and regulatory guardrails.