Microsoft has started integrating Anthropic’s Claude models into its flagship Copilot platform. For years, Microsoft’s AI identity was closely tied to OpenAI.

This integration marks a strategic broadening rather than a split. Microsoft invested billions in OpenAI, embedding ChatGPT-like capabilities across products from Bing to Office.

While that partnership is far from over, the latest update reveals a more complex approach: Copilot will no longer rely on a single AI model.

Anthropic’s Claude Sonnet 4 and Claude Opus 4.1 will run alongside OpenAI’s models, offering businesses greater choice.

Copilot is evolving from a branded OpenAI product into a multi-model operating system for work.

Why this matters

The change is about more than just new features; it reflects considerations of control, risk, and leverage.

Control: Relying solely on OpenAI risked tying Microsoft to one vendor’s product roadmap. Adding Anthropic reduces this dependency.

Risk: AI models can exhibit quirks, hallucinations, or ethical blind spots. Using multiple models diversifies capabilities and lessens these risks.

Leverage: A broader AI lineup enhances Microsoft’s negotiating power with OpenAI and potential partners.

The Claude factor

Anthropic is known for prioritising safety, with strong guardrails and reliability. For organisations cautious about AI unpredictability in sensitive tasks, Claude offers a compelling alternative.

However, Anthropic’s models will be hosted outside Microsoft’s cloud, on Amazon Web Services. This signals a new era of multi-cloud AI, where Microsoft relies on rivals to meet customer needs.

For OpenAI, this move underscores that Silicon Valley partnerships are never guaranteed. For Anthropic, it’s a major win, embedding its brand into millions of Microsoft 365 workflows.