Two of the biggest names in technology are working to reduce their reliance on Anthropic's Claude, according to a report from The Information. Meta and Microsoft have both reportedly been taking steps to move away from Claude and toward alternative AI models, a development that raises questions about the long-term business relationship between Anthropic and its largest enterprise partners.
What the Shift Means for Anthropic
The report arrives at a sensitive moment for Anthropic. The company has aggressively expanded its enterprise relationships over the past year, positioning Claude as a go-to model for businesses that want a safety-focused alternative to OpenAI's offerings. Losing or seeing reduced usage from customers the scale of Meta and Microsoft would be a meaningful blow to that strategy, even if the full financial impact remains unclear.
Microsoft's reported effort to pull back from Claude fits with earlier signals from inside the company. A memo that circulated internally suggested there was growing friction around Claude's role in Microsoft's product stack, pointing to potential trouble ahead for the partnership.
Key Facts
- Meta and Microsoft are both reportedly working to reduce their use of Claude, according to The Information.
- The moves suggest big tech companies want more control over their AI dependencies.
- Anthropic has invested heavily in enterprise and developer partnerships to grow Claude's adoption.
- Microsoft has its own deep investment in OpenAI, giving it an obvious in-house alternative to Claude.
- Meta has been building out its own Llama model family, reducing its need for third-party models.
For Microsoft, the path away from Claude is relatively straightforward. The company holds a multibillion-dollar stake in OpenAI and has integrated GPT models across its product lines, from Copilot to Azure AI services. Using Claude required going outside that existing relationship, and pulling back simply means consolidating around what Microsoft already has. The dynamic for Meta is different but equally logical. Meta has invested heavily in its open-source Llama model family and has strong reasons to build internal capability rather than rely on an external API.
The broader trend here is that large technology companies are treating AI model access the way they once treated cloud infrastructure: as something they eventually want to own or control, not just rent.Industry analysis, ClaudeAINews.com
Anthropic's Position in a Shifting Market
Anthropic has been working to diversify how Claude reaches customers. Products like Claude Docs represent a push into end-user applications rather than just API access, which could provide more insulation from the kind of enterprise pullback that Meta and Microsoft reportedly represent. The company has also been building out deeper integrations with workplace tools, including a Claude agent designed for Microsoft Teams, which reflects an attempt to get Claude embedded at the workflow level where switching costs are higher.
The competitive picture is also shifting at the model level. Anthropic has been active in launching new versions of Claude and competing directly with OpenAI, Google, and Meta's own releases. That pace of model development matters because enterprise customers tend to stick with whatever model performs best on their specific tasks. If Claude maintains a performance edge, some of the commercial pressure may ease. Still, the news that two major partners are actively looking to move on is not something Anthropic can easily dismiss.
Investors and observers will be watching how Anthropic responds. The company has raised billions of dollars and carries a valuation that assumes continued growth in enterprise adoption. A sustained pullback from large customers would put pressure on that assumption. For now, the details of any formal changes to Meta's or Microsoft's agreements with Anthropic have not been disclosed, and neither company has issued a public statement on the matter. The story continues to develop, and further details are likely to emerge as both companies make their AI sourcing decisions more visible through product announcements and procurement moves.
“When big tech builds its own AI muscle instead of licensing yours, the message to every organisation is clear: don't architect your workflows around a single third-party model, because today's preferred provider is tomorrow's redundant dependency.”
Leon Tindemans, AI expert and entrepreneur specialising in Claude, Copilot and ChatGPT. Learn more with Copilot training by TTM Communicatie.
