Stock Markets May 28, 2026 10:18 AM

Microsoft Accelerates Shift to Internal AI Models as Shares Jump 3%

Company to unveil a new coding model and broader in-house AI lineup at Build as it seeks to reduce reliance on external providers

By Marcus Reed MSFT

Microsoft shares climbed 3% on Thursday after reports that the company will disclose a suite of internal AI models at its Build developer conference in San Francisco. The rollout, which includes a coding model aimed at enhancing GitHub Copilot, is part of an effort to cut dependence on external suppliers such as OpenAI and Anthropic and to lower the costs of embedding AI into Microsoft products.

Microsoft Accelerates Shift to Internal AI Models as Shares Jump 3%
MSFT

Key Points

  • Microsoft shares rose 3% on Thursday following reports of a planned reveal of new in-house AI models at the Build conference in San Francisco.
  • The announced suite is said to include a coding model for GitHub Copilot and models focused on reasoning, speech, transcription, and image generation.
  • Microsoft currently uses models from OpenAI (provided free until 2032) and Anthropic; the new effort is intended to reduce reliance on those external suppliers and lower AI operating costs, including within Office applications.

Microsoft (NASDAQ:MSFT) shares rose 3% on Thursday after reports indicated the company plans to introduce a set of internally developed AI models at its Build developer conference in San Francisco.

According to the report, the new lineup will include a coding model intended to strengthen GitHub Copilot's competitiveness with rivals such as Cursor and Claude Code. In addition to the coding model, Microsoft is expected to unveil models focused on reasoning, speech, transcription, and image generation.

The move comes as Microsoft continues to rely heavily on external providers for AI capabilities. The company currently uses models from OpenAI - provided at no cost under an agreement that runs through 2032 - as well as models from Anthropic to power parts of its software. The reported strategy aims to build in-house alternatives that can match or approach the capabilities of those external models while offering lower operational costs.

Internal sources cited in the report note that Microsoft’s internal AI group, led by AI chief Mustafa Suleyman, had faced limits on training top-tier models because of terms in the OpenAI agreement. Those restrictions were renegotiated in April, the report said, clearing the way for development of the new models.

Microsoft plans to position the internally built models as more economical options compared with slightly more capable models from OpenAI and Anthropic. The company expects that using its own models could reduce the expense of running AI-powered features, including those integrated into its Office applications.

Executives and engineers at Microsoft are reportedly preparing to present the new models to developers next week at Build. The product lineup is framed as part of a broader effort to decrease dependence on external AI suppliers once preferential arrangements with those suppliers expire.


Context and implications

The reported initiative represents a push by Microsoft to internalize AI capabilities and manage future costs linked to AI services embedded across its software portfolio. Details published ahead of Build indicate the company will emphasize cost efficiency and competitive parity with third-party providers.

Risks

  • Microsoft remains dependent on external AI providers today - any delays or shortcomings in developing competitive in-house models could prolong reliance on OpenAI and Anthropic. This impacts cloud software and enterprise applications.
  • Terms of the OpenAI agreement previously limited internal training of top-tier models; while restrictions were renegotiated in April, the effectiveness and timing of internal development remain uncertain. This uncertainty affects product roadmaps for developer tools and productivity suites.
  • Positioning in-house models as cheaper alternatives to slightly more capable external models may create trade-offs between cost savings and model performance, with implications for user experience in productivity and developer tools.

More from Stock Markets

Toronto market ends at fresh record as healthcare, financials and materials lead gains Jun 4, 2026 After-Hours Movers: Lululemon Dips on Guidance as Software and Data Names Show Mixed Reactions Jun 4, 2026 Anthropic Places Engineers Inside NSA to Support Mythos AI for Offensive Cyber Tasks Jun 4, 2026 Trump Directs $700M Toward Coal Industry, Lifting Peabody Shares Jun 4, 2026 Pentagon Poised to Abandon Tomahawk Deployment to Germany Over Escalation Concerns Jun 4, 2026