Like two star-crossed lovers, Microsoft and Anthropic are becoming more deeply entwined.
Don’t worry, this isn’t romantic fiction, but the world of business. The two firms have announced a major expansion of their partnership, introducing public preview access to several Claude 4.5 and 4.1 models inside Microsoft Foundry and deepening integrations within Microsoft 365 Copilot.
The move broadens the availability of Anthropic’s AI models for enterprise developers and signals Microsoft’s continued push to diversify its AI offerings beyond its existing OpenAI-powered stack.
Claude models land in Microsoft Foundry
Under the expanded partnership, Claude Sonnet 4.5, Haiku 4.5, and Opus 4.1 are now accessible to Azure customers through Microsoft Foundry, the company’s platform for building production-grade AI systems and agent workflows.
According to the announcement, this gives organizations the ability to build with “Claude, the world’s best models for coding, agents, and office tasks,” while staying inside their existing Microsoft infrastructure. Because the models are offered serverlessly through Foundry, Anthropic manages the infrastructure and scaling, enabling faster deployment for enterprises that want to experiment or move straight to production.
Crucially, the models are eligible for organizations’ Microsoft Azure Consumption Commitment (MACC). This means customers who have already negotiated spending commitments with Microsoft can adopt Anthropic’s models without opening a separate vendor contract. For large enterprises, this eliminates procurement delays that often stretch into months.
Deeper integration with Microsoft 365 Copilot
The partnership also strengthens Claude’s footprint in Microsoft’s productivity suite. Claude already powers the Researcher agent within Microsoft 365 Copilot, handling complex multi-step research tasks and supporting custom agent development in Copilot Studio.
Now, Excel is joining the lineup. Microsoft’s new Agent Mode in Excel includes an option to use Claude to “build and edit spreadsheets directly,” including generating formulas, analyzing data, identifying errors, and iterating on solutions. This integration positions Claude as a direct assistant inside one of the most widely used enterprise software tools globally.
The implications are significant: enterprises that were previously hesitant to adopt standalone AI solutions can now access advanced generative models directly inside their core productivity workflows, with no new procurement or IT overhead.
Choosing the right Claude model
Microsoft customers can now select between several Claude models depending on the complexity, speed, and cost needs of their applications:
Sonnet 4.5 is optimized for multi-step reasoning, agentic workflows, and autonomous coding tasks. Its arrival inside Foundry positions it as a competitor to OpenAI’s o-series models for enterprise coding automation.
Haiku 4.5 offers near-frontier performance at roughly one-third the price of Sonnet, making it ideal for high-volume applications. Use cases include customer support, content moderation, real-time coding assistance, and sub-agent orchestration. For companies building large-scale automation systems, Haiku’s lower cost could materially reduce operating expenses.
Opus 4.1 is positioned as a specialized reasoning model, and it handles extended, complex problem-solving and long-form analysis. Its inclusion strengthens Anthropic’s portfolio for enterprises requiring precise, technical reasoning — common in fields such as finance, legal research, and scientific analysis.
All models in Foundry support advanced capabilities from the Claude Developer Platform: tool use, vision, web search and fetch, citations, code execution, and prompt caching. This means developers can build full-stack agents that retrieve data, reason through problems, and act autonomously inside enterprise environments.
Enterprise implications
For organizations already invested in Microsoft Foundry or Microsoft 365 Copilot, this partnership removes one of the biggest historical blockers to adopting new AI tools: vendor complexity. Historically, adding a new model or provider required negotiating separate contracts, gaining procurement approval, and reconciling new billing cycles.
By integrating Anthropic directly into Microsoft’s pricing and billing ecosystem, the partnership effectively turns Claude into an “add-on” within Azure — lowering the barrier to experimentation and adoption. This could accelerate multi-model strategies among enterprises, where organizations rely on different AI systems optimized for different tasks rather than a single general-purpose model.
Sundar Pichai has described the AI boom as an “extraordinary moment” but acknowledged “elements of irrationality” emerging as valuations skyrocket and capital floods into the sector.