Microsoft to Replace OpenAI Image Models With Cheaper In-House Technology

Microsoft is replacing OpenAI's image-generating models with its own in-house technology across major products including Bing and PowerPoint. The strategic shift is driven by a push for greater efficiency and significant cost reductions in deploying artificial intelligence, according to the company's AI chief.

A Strategic Shift to In-House AI

In an interview, Microsoft AI CEO Mustafa Suleyman confirmed the transition, positioning it as part of the company's broader effort to develop proprietary AI models that compete with those from industry leaders like OpenAI and Anthropic. This move allows Microsoft greater control over its AI development and deployment pipeline.

By integrating its own technology, Microsoft aims to create a more seamless and optimized experience within its product ecosystem. Suleyman noted that image creation is a critical function and often one of the first AI prompts new users try, making a high-quality, fast experience essential for user retention.

Focus on Cost and Performance

The primary driver for the change is economic efficiency. According to Suleyman, Microsoft's proprietary models deliver substantial savings and performance improvements. Key benefits cited include:

Ad

• Cost Reduction: The in-house models cost approximately 85% less to run in PowerPoint compared to OpenAI's technology.
• Performance: Suleyman described the Microsoft models as "faster, it’s cheaper, it’s higher quality, it drives better retention."

While Microsoft's partnership provides it with access to OpenAI models, the company must still cover the significant computing infrastructure costs needed to run them. Utilizing more efficient, internally developed alternatives allows Microsoft to realize substantial operational savings.

Implications for the AI Landscape

This decision highlights Microsoft's growing confidence in its own AI capabilities and its strategy to reduce operational dependency on partners, even one as crucial as OpenAI. It underscores the intense industry-wide focus on model efficiency, as the high cost of computing power remains a major factor in the profitable deployment of AI services at scale. The move signals a maturing of Microsoft's AI stack as it seeks to compete on both performance and cost.