🔍 Read the full analysis: Meta And Microsoft’s Claude Pullback Highlights The Complexity Of Switching on ThorstenMeyerAI.com
Get office and shipping supplies 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
The Information reported on Oct. 5 that Meta and Microsoft have reduced or lowered projected internal use of Anthropic’s Claude tools while directing employees toward alternatives they own or back. The report attributes the moves to costs and spending controls, not a stated judgment that Claude performs worse. The companies’ ability to switch reflects their existing tools and engineering resources—conditions many businesses may not share.
Meta and Microsoft are steering some employees toward alternatives to Anthropic’s Claude tools, according to an Oct. 5 report by The Information, which cited internal usage and spending changes. The reported moves concern the companies’ own workforces, not a full withdrawal of Claude from their products, and highlight how difficult switching AI systems can be for businesses without ready-made substitutes.
The Information reported that Meta’s Claude Code use fell from about 60,000 employees earlier this year to about 30,000. The company has directed staff toward its own coding tools, MetaCode, which the source material says has more than 30,000 internal users, and Muse Code, with more than 6,000. The figures describe reported internal users; they do not establish how often each tool is used or whether the users overlap.
At Microsoft, the report said the company had projected more than $1 billion a year in internal spending on Anthropic technology, including Claude Code, Claude models in Copilot and Claude Mythos. Microsoft reportedly lowered that projection by more than a third and has been steering employees toward GitHub Copilot and OpenAI models. The source material also cites a report that some monthly team budgets fell from roughly $100,000 to roughly $10,000; that detail is attributed to a single account and has not been independently established here.
The stated reasons are cost and spending controls, alongside efforts to promote tools the companies own or support. Neither company is reported as saying Claude performed worse. The reported changes also do not mean Claude access has ended: 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.
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.
Why Internal Switching Is Hard
The moves matter because they show that large AI buyers can shift workloads when they already have alternatives in place. Meta and Microsoft have their own or closely aligned coding products, substantial engineering teams and the scale to justify the expense of moving work. Their decisions are not a simple test of which model is better; they also reflect supplier costs, internal priorities and competitive interests.
For other organizations, a lower model price does not automatically mean a lower total cost. Teams may need to repeat evaluations, adjust prompts and tool connections, and retrain employees. Coding systems also depend on integration with editors, repositories and team practices. A switch can disrupt that fit, while changing providers may affect cached context and the economics of repeated agent tasks.
Quality differences can carry costs that are harder to see on an invoice. A model that performs less well on a company’s tasks may require extra human review, rework or correction of errors. Comparing token prices alone can miss those expenses. The reported corporate moves offer evidence that switching is possible at scale for these firms, but not that it will save money for every buyer.
As an affiliate, we earn on qualifying purchases.
What the Report Covers
The report concerns internal employee use and spending projections, not a general decision by Meta or Microsoft to stop offering or buying Claude. Microsoft’s use of Anthropic models in customer-facing Copilot features is reported to continue. Those distinctions limit what can be concluded from the reported reductions in internal use.
Both companies also have reasons to favor alternatives beyond model performance. Meta develops its own models and coding tools. Microsoft owns GitHub Copilot and is a major backer of OpenAI. Steering employees toward products connected to their own businesses can reflect ordinary commercial incentives as well as cost management. The report does not establish that either company found Claude inferior.
The source material argues that businesses should be able to route different tasks to different models rather than depend on a single provider. That approach can preserve options when prices, policies or availability change. But having more than one model available is not the same as being ready to switch: a second system needs real integrations, tested workflows and clear measures of success.
What the Figures Cannot Show
The reported numbers do not reveal how much work moved, how usage was counted, or whether the same employees used multiple tools. The Microsoft spending figure is described as a projection, not confirmed annual expenditure, and the report does not provide enough detail here to calculate realized savings. The team-budget figures are also based on a single account.
It is unclear how the changes affected productivity, output quality or review time, and whether the alternatives perform as well on the companies’ specific tasks. The reporting cited in the source material does not give a detailed breakdown by model, workload or business unit. It also does not establish whether the internal changes will be permanent. No reported statement identifies model quality as the reason for the pullback.
Evidence Buyers Should Watch
Further reporting or statements from Meta, Microsoft and Anthropic may clarify the scope of the changes, the amount of spending involved and whether customer-facing use is changing. The key evidence will be more than employee counts or budgets: buyers will need data on completed tasks, output quality, review effort and total cost across comparable workloads.
For organizations weighing their own options, the reported case points to practical preparation rather than an automatic decision to switch. Teams can test another model on a limited share of real work, maintain representative evaluation tasks and keep prompts and business logic portable. Those measures can reduce the cost of a future move, but they require engineering and measurement before a supplier change becomes urgent.
Until fuller performance and spending data are available, the report supports a narrower conclusion: Meta and Microsoft are reported to be redirecting some internal work toward alternatives amid cost controls, and their existing tools make that choice more feasible. Whether the same approach pays off for other companies depends on their workloads, integration costs and results.
Key Questions
Have Meta and Microsoft stopped using Claude?
No full withdrawal is reported. The report describes changes to internal employee use and spending projections. The source material says Microsoft continues to use Anthropic models in customer-facing Copilot features.
Why are the companies steering employees toward other tools?
The reported reasons include rising costs, tighter spending controls and a preference for tools they own or support. Neither company is reported to have said Claude performed worse.
What did the report say about Meta’s Claude Code use?
The Information reported that employee use fell from about 60,000 earlier this year to about 30,000. It also reported more than 30,000 internal users of MetaCode and more than 6,000 of Muse Code.
Does switching models always save money?
No. A lower model bill can be offset by engineering work, repeated evaluations, integration changes, retraining and extra review or rework. The report does not establish that switching will save money for other companies.
What is still unknown about the reported changes?
The public information summarized here does not show the full scale of moved workloads, realized savings or effects on productivity and quality. The Microsoft spending figure is a projection, and the duration of the internal changes is unclear.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
