Microsoft is stepping up its artificial intelligence efforts by committing to major investments in developing and training its own frontier models. This move adds another layer to the company’s already complex partnership with OpenAI, while signaling its ambition to compete directly with leading AI players like Google, Meta, and xAI.

Last month, Microsoft AI unveiled its first in-house models, marking the beginning of a broader strategy. At a company town hall on Thursday, Microsoft AI chief Mustafa Suleyman revealed that the tech giant is scaling up its infrastructure. “We should have the capacity to build world class frontier models in house of all sizes, but we should be very pragmatic and use other models where we need to,” he said. Suleyman also noted that the company is investing heavily in its compute clusters, pointing out that the MAI-1-preview was trained on 15,000 H100s, which he described as “a tiny cluster in the grand scheme of things.”

Suleyman further hinted at larger ambitions, suggesting that future models may be trained on clusters “six to ten times larger” than the current setup. “Much more to do, but it’s good to take the first steps,” he added.

Microsoft CEO Satya Nadella echoed the vision, stating he is “looking forward to us building model capability, so that we can build model-forward products.” Nadella reaffirmed Microsoft’s commitment to supporting multiple models, citing GitHub Copilot as a prime example.

Meanwhile, reports suggest that Microsoft 365 Copilot may soon integrate Anthropic’s AI models, which reportedly outperformed OpenAI in tools like Excel and PowerPoint.