🔍 Read the full analysis: What It Takes To Switch From Claude, As Meta And Microsoft Pull Back on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
TL;DR
Meta and Microsoft have reportedly reduced employees’ use of Anthropic’s Claude tools and redirected some work to products they own or already use. The reported moves reflect spending controls and available substitutes, not a stated finding that Claude performs worse. For other companies, changing models can require substantial work on evaluations, integrations, training and quality checks.
Meta and Microsoft have reportedly reduced internal use of Anthropic’s Claude tools, directing some employees toward their own products and other alternatives, according to a report by The Information published October 5. The reported decisions matter beyond the companies’ spending: they show how difficult model switching can be for organizations without established alternatives, and do not amount to evidence that Claude performed worse.
The Information reported that Meta cut Claude Code use among employees from about 60,000 earlier this year to about 30,000. The company has been steering staff toward its internal coding tools, MetaCode, which the source report says has more than 30,000 internal users, and Muse Code, with more than 6,000. Those figures describe reported internal adoption, not customer usage.
Microsoft had reportedly projected more than $1 billion a year in internal spending on Anthropic technology, including Claude Code, Claude models used in Copilot and Claude Mythos. The report says Microsoft subsequently reduced that projection by more than a third and has directed employees toward GitHub Copilot and OpenAI models. It also says some internal team token budgets were tightened; a reported example of monthly budgets falling from roughly $100,000 to $10,000 comes from a single account and has not been independently established here.
The reported changes concern employees’ tools and projected internal spending. The source material says Microsoft continues to use Anthropic models in customer-facing Copilot features and that customer spending on Claude through Microsoft platforms is growing. Neither company is reported to have ended access to Claude. The reported reasons for the internal shift include cost, spending controls and the availability of tools the companies own or support.
Meta and Microsoft pulled back from Claude. Here’s what switching actually costs.
The Information reports both companies steering their own employees away from Claude. Read as a verdict on Claude, it misleads. Read as a demonstration of switching — and who can afford it — it’s the most useful enterprise-AI signal this month.
Staff steered to GitHub Copilot and OpenAI models; stricter token budgets. One unconfirmed report: some team budgets ~$100k → ~$10k/month.
Microsoft reportedly still spends heavily on Claude for customer-facing Copilot — and that spending is reported to be growing.
Reported drivers: rising token costs and owned alternatives. Neither company is reported to have called Claude worse.
Meta builds coding tools; Microsoft owns Copilot and backs OpenAI. This is ordinary vertical integration.
Keep a second vendor live on real work.
A few hundred tasks with pass criteria.
Logic, prompts, tools in your layer.
Tokens are the cheap half.
Know what you’d rebuild.
On the evidence reported, Meta and Microsoft didn’t reject Claude. They brought spending in-house where they could and kept buying where they couldn’t — Microsoft remains a large Anthropic customer for the products it sells. The signal is the mechanism: the most sophisticated buyers treat models as interchangeable suppliers behind a layer they control.Meta could halve its Claude usage because it had built somewhere else to go. Build somewhere else to go.
Switching Depends on Existing Alternatives
For organizations buying AI tools, the report highlights a difference between being able to choose another model and being able to move work to it without disrupting operations. Meta and Microsoft have products and engineering capacity available as substitutes. A company that relies on one provider may first need to build or validate those alternatives, making a quick response to higher prices or changed terms much harder.
Switching costs can extend beyond a new subscription or token rate. Teams may need to rerun evaluations, adjust prompts and agent tools, reconnect software integrations and retrain employees. A different model can also change output quality on a company’s specific tasks, potentially increasing review, rework or errors. Those effects are harder to capture than a bill and can outweigh apparent savings.
The financial scale reported for Microsoft illustrates why large buyers may accept the effort. A cut of more than a third from a projection above $1 billion would imply a reduction exceeding $300 million if applied to that full projection. That is an arithmetic implication, not a confirmed realized saving: the report concerns a spending projection, and the actual amount saved is not stated. The calculation also says little about what a smaller organization might save after accounting for migration costs.
As an affiliate, we earn on qualifying purchases.
Internal Use Is Not Customer Access
The reported development is about how two companies direct their own employees, not a general withdrawal of Claude from the market. Microsoft’s continued reported use of Anthropic models in customer-facing Copilot features is a key distinction. Internal procurement choices and products offered to customers can follow different commercial and technical needs.
Both companies also have alternatives tied to their own businesses: Meta develops its own models and coding tools, while Microsoft owns GitHub Copilot and backs OpenAI. That gives them options that many customers do not have. Their decisions may reflect a combination of spending discipline and strategic preference; the reporting does not isolate how much each factor contributed.
The source material argues that enterprises should make it possible to route work among more than one model. In practice, this means maintaining more than a vendor API connection: teams need tested workflows, usable evaluation data and a way to compare the results and total costs. The reported moves offer an example of large companies shifting internal work, but do not establish that every business can reproduce the same savings.
As an affiliate, we earn on qualifying purchases.
Savings and Quality Effects Remain Unclear
The reports do not provide audited figures for actual spending or realized savings, nor a full breakdown of how the companies calculated their revised projections. The number of employees using a tool does not show how often they use it, how much work they complete with it or whether usage has shifted permanently.
It is also unclear how much of the reported change came from token prices, tighter budgets, internal product strategy or other factors. The source material does not report that Meta or Microsoft judged Claude to be lower quality. It gives no direct comparison of performance, review time or productivity across the tools, and the budget reduction example is based on a single account.
For companies considering a similar move, the migration costs will vary by workflow, existing integrations and the quality of available alternatives. No general savings estimate can be drawn from the reported figures alone.
As an affiliate, we earn on qualifying purchases.
Track Adoption, Costs and Output
The next evidence to watch is whether Meta and Microsoft report further changes to internal tool use or spending, and whether Microsoft’s customer-facing use of Anthropic models continues alongside the internal shift. The report does not set out a timetable for either company’s plans.
Businesses weighing their own options can compare models on representative tasks and accepted results, rather than token prices alone. That requires tracking review and rework, checking integrations and giving employees time to adapt. Keeping a second provider active on a limited share of real work can make future comparisons more practical, but it also has costs and does not guarantee that switching will pay off.
AI model training and quality check software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
Have Meta and Microsoft stopped using Claude?
No such company-wide stop is reported. The report concerns reduced internal use and redirected employee work. The source material says Microsoft continues to use Anthropic models in customer-facing Copilot features.
Why are the companies reportedly steering employees away from Claude?
The reported reasons include cost, tighter spending controls and the availability of alternatives. The report does not say either company found that Claude performed worse.
How much could Microsoft save?
The source material says Microsoft cut a projected annual internal spend of more than $1 billion by more than a third. That implies a reduction exceeding $300 million against the projection if the calculation applies to the full amount, but it is not a confirmed figure for realized savings.
Why can switching models be difficult for other companies?
A move can require new evaluations, prompt and tool changes, integration work and employee training. Differences in model quality may also affect review time and rework, so a lower token bill does not by itself establish lower overall cost.
Does this report show that Claude is losing customers?
It documents reported changes to internal use at two companies. It does not establish a broader loss of Claude customers, and the source material says customer spending on Claude through Microsoft’s platforms is growing.
Source: ThorstenMeyerAI.com
Evergreen bestsellers Picks
bestsellers
As an affiliate, we earn on qualifying purchases.
