Two of the world's biggest technology companies are quietly pulling back from a leading rival's artificial intelligence assistant, a sign that the battle for enterprise AI is increasingly fought over which tools employees use by default.
According to a report from The Information, Meta and Microsoft are both steering their staff away from Anthropic's Claude and toward their own in-house AI coding tools. At Microsoft, the shift is showing up in the budget: the company reportedly cut its projected internal spending on Claude by more than a third as it pushes developers to use GitHub Copilot first. At Meta, the change is visible in headcount: the number of employees using Claude Code reportedly fell to about 30,000 from roughly 60,000 earlier this year, as the company promotes internal options like Muse Code and MetaCode.
The common thread is control. When a company makes its own assistant the default place to write prompts and generate code, it can manage costs, standardize which models are allowed, and limit what sensitive data leaves its systems. That can shrink paid “seats” and API calls for outside model providers, even if some teams prefer them.
Why big tech is favoring in-house AI
For Microsoft, the decision to prioritize GitHub Copilot is not surprising. Copilot is deeply integrated into its developer tools and cloud services, and the company has been turning Copilot into an always-on digital coworker, part of a broader push to make it the default assistant across its ecosystem. By routing employees through its own product, Microsoft not only saves money but also gathers usage data that can improve its own models and reinforce its platform lock-in.
Meta, meanwhile, has invested heavily in building its own AI infrastructure and models, including the open-source Llama family. Encouraging employees to use Muse Code and MetaCode helps the company test and refine its internal tools, while also keeping proprietary code and data within its own walls. For a company that handles vast amounts of user data, limiting exposure to third-party AI services is a security and compliance advantage.
This is not just a cost-saving exercise. It reflects a broader trend in enterprise AI: the model that wins is often the one that is embedded in the default workflow, not necessarily the one that performs best in benchmarks. As Google's Gemini 4 Argon rollout showed, even top-tier models can face real-world coding gaps that make companies cautious about switching defaults.
What it means for Anthropic and the AI market
For Anthropic, the news is a reminder that winning enterprise customers is about more than having a capable model. It needs to get formally approved and embedded inside customers' internal tooling, where pricing power and growth depend on being part of the default workflow. The company has been preparing for an IPO, and launching Claude Sonnet 5.5 at the same price suggests it is trying to hold its ground in a competitive market.
But if two of the largest potential customers are reducing their reliance on Claude, that could weigh on Anthropic's revenue growth and its ability to command premium pricing. The company still has a strong position in many enterprises, but the shift at Microsoft and Meta highlights the risk that even the best third-party model can be displaced by a company's own tools.
For investors, the story is a useful reminder that enterprise AI adoption is not just about which model is best; it is also about procurement and platform “defaults.” If Microsoft and Meta route employees through GitHub Copilot or their own coding stacks, they effectively decide which vendors get usage, data, and recurring revenue. That dynamic could shape the fortunes of AI companies like Anthropic, as well as the broader AI rally that has lifted Wall Street.
What investors should watch
For everyday investors, the key takeaway is that the AI boom is not just about who builds the best model. It is about who controls the distribution channels and default settings inside corporate IT departments. Companies that own the platform—like Microsoft with GitHub Copilot and Meta with its internal tools—have a structural advantage in steering usage toward their own products.
That could mean more volatility for pure-play AI vendors like Anthropic, especially as it heads toward a public listing. It also suggests that the biggest winners in AI may be the companies that already have deep distribution, not just the ones with the most advanced research.
As always, investors should keep an eye on how these dynamics play out in earnings calls and product announcements. The next few quarters will show whether other large enterprises follow Microsoft and Meta's lead, and whether Anthropic can adapt by embedding itself more deeply into customers' workflows.
