Microsoft is reportedly shifting its AI strategy, signaling a gradual move away from relying solely on OpenAI’s models. Currently, Microsoft’s AI ecosystem focuses heavily on ChatGPT, DALL-E 3, and other OpenAI technologies, powering tools like Microsoft 365 Copilot and GitHub Copilot. While enterprise adoption has been strong, consumer-facing AI initiatives did not take off.
Mustafa Suleyman, Microsoft’s AI lead and co-founder of DeepMind, told the Financial Times that Microsoft plans to develop its own “frontier” AI models using massive compute resources and top-tier training teams. The goal is to build homegrown models for tasks where Microsoft wants full control, even as OpenAI continues to play a role in specific applications.
The announcement follows a restructuring of Microsoft’s deal with OpenAI last year, which allowed OpenAI to seek compute elsewhere and reduced Microsoft’s financial exposure to the company’s soaring operational costs. OpenAI has struggled to turn a profit, with projected compute contracts exceeding a trillion dollars and ongoing dependence on outside investors.
Despite this shift, Microsoft insists the partnership isn’t over. Frank X. Shaw, Microsoft’s communications chief, emphasized a “multi-model world” where OpenAI remains part of the ecosystem while Microsoft develops its own cutting-edge systems. The company is already investing in other AI startups like Anthropic and plans to launch new models sometime in 2026.
Suleyman painted an optimistic vision for AI, highlighting its potential to improve healthcare and automate workflows while ensuring these systems remain subordinate to human oversight. At the same time, he acknowledged the societal and ethical concerns surrounding AI, including job disruption and autonomous decision-making, stressing that Microsoft’s tools are meant to enhance human well-being rather than replace it.