Microsoft Challenges OpenAI and Anthropic With AI Strategy

Microsoft’s AI strategy now centers on giving enterprises access to multiple models while keeping the company’s cloud, agent, security, and data platforms at the heart of their operations.

The approach also reflects Microsoft’s increasingly complicated relationships with OpenAI and Anthropic. Microsoft remains a major investor and partner, but both AI companies are expanding into enterprise applications, agents, and infrastructure that compete directly with its products, as TechCrunch reports.

Microsoft has strong financial reasons to protect its position. The company recently reported $90 billion in quarterly revenue and $35.8 billion in net income, driven partly by demand for cloud and AI services.

Nadella wants enterprises to avoid model lock-in

Microsoft CEO Satya Nadella is encouraging companies to use several AI models instead of depending on a single provider.

His strategy separates the agent framework, enterprise data, and security controls from the model handling each task. Businesses can then replace models based on cost, performance, reliability, privacy, or regulatory requirements.

Under this approach, Microsoft provides the infrastructure that connects models, agents, business applications, security systems, and company data. The individual AI model becomes a replaceable part of a broader enterprise platform.

Nadella argues that relying on one frontier model creates several risks. Companies may need to send sensitive internal information to an external provider, while model changes, outages, refusals, or new restrictions could interrupt important operations.

Hugging Face incident supports Microsoft’s argument

Nadella cited a recent Hugging Face security incident to explain why companies may need access to competing models.

Hugging Face reportedly attempted to use a private AI model to investigate the incident, but the model refused to assist. The company then turned to the open-source Z.ai GLM 5.2 model to analyze logs and help defend its systems.

The incident supports Microsoft’s argument that an organization may need one model to investigate or contain problems linked to another. A multi-model setup also gives businesses another option when a provider changes its policies or cannot complete a task.

Azure becomes Microsoft’s model control layer

Azure now offers access to more than 11,000 AI models, including systems from OpenAI, Anthropic, Mistral, xAI, and Microsoft’s own MAI family.

Microsoft says customers can choose models based on quality, speed, price, security, and compliance. The company still supports frontier models, but it does not want customers to build their entire AI architecture around one provider.

Microsoft is taking a similar approach to hardware. Azure will use both AMD Helios and NVIDIA Vera Rubin systems, reducing its dependence on one chip supplier.

This strategy lets Microsoft benefit regardless of which model or hardware platform an enterprise chooses, provided the customer continues using Azure and Microsoft’s management tools.

Microsoft accelerates its own MAI models

Microsoft is also developing more of its own AI technology. Nadella said the company is accelerating work on the MAI model family, which now covers image generation, voice, transcription, coding, reasoning, and cybersecurity.

One recent release, MAI-Cyber-1-Flash, targets security analysis and automated cyber defense.

Microsoft is designing these models alongside its Maia AI processors. The company claims its MAI models deliver 40% better performance per watt when running on Maia 200 hardware.

It is presenting the combination as a more efficient and affordable alternative to competing AI systems. Microsoft also claims MAI-Cyber-1-Flash can outperform Anthropic’s much larger Mythos security model and operate at half the cost when connected to Microsoft’s multi-agent security platform.

Microsoft’s partners are becoming direct competitors

OpenAI and Anthropic remain important Microsoft partners, and their growth continues to benefit the company financially. Anthropic recently delivered one of Microsoft’s largest quarterly investment gains.

However, both companies are building agents, enterprise applications, coding tools, and infrastructure that could weaken Microsoft’s control over customer relationships.

Microsoft’s response is not to abandon frontier models. Instead, it wants businesses to use them through a platform that Microsoft controls.

The company ultimately aims to manage the cloud infrastructure, agent framework, security layer, model marketplace, enterprise data, and customer relationship, even when another company supplies the underlying AI model.