🔍 Read the full analysis: Meta And Microsoft Pulled Back From Claude. Here’s What Switching Actually Costs. on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
The Information reported on Oct. 5 that Meta and Microsoft are reducing employees’ use of Anthropic’s Claude tools and directing more work to alternatives they control or back. The reported shift concerns internal use, not an end to Claude access or a broad rejection of Claude for customers. It also highlights the engineering, evaluation and productivity costs that can make switching difficult for companies without ready substitutes.
Meta and Microsoft are steering employees away from some Anthropic tools and toward alternatives they own or back, according to an Oct. 5 report by The Information. The reported changes concern internal use, and do not establish that either company has ended Claude access or stopped using Anthropic technology for customer-facing products.
The Information reported that Meta reduced the number of employees using Claude Code from about 60,000 earlier this year to about 30,000. The report says Meta is directing staff toward its own coding tools: MetaCode, which has more than 30,000 internal users, and Muse Code, which has more than 6,000. These figures describe reported internal users, not a comparison of product performance or customer adoption.
For Microsoft, the report described a cut of more than a third to a prior projection of over $1 billion a year in internal Anthropic-related spending. That projected spending reportedly included Claude Code, Claude models in Copilot and Claude Mythos. Microsoft is steering employees toward GitHub Copilot and OpenAI models, according to the account. The report also says Microsoft continues to spend on Anthropic models for customer-facing Copilot features and that customer spending on Claude through Microsoft platforms is growing.
The reported reasons for the internal changes include rising token costs and tighter spending controls, alongside a preference for tools the companies own or are invested in. The source material does not report that Meta or Microsoft said Claude performed worse. A separate detail in the report says some Microsoft team budgets may have fallen from around $100,000 a month to around $10,000; that figure is attributed to a single account and should not be treated as a company-wide budget rule.
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 Ready Alternatives Matter
The reported decisions matter less as a verdict on one model than as an example of how large buyers can shift AI workloads. Meta and Microsoft have alternatives already in use, giving them options when costs, budgets or company priorities change. The report does not establish that the alternatives are better; it shows that the companies can direct some internal work elsewhere.
For other businesses, the key question is whether a model can be replaced without disrupting work. A lower token price does not by itself show that a switch saves money. Companies may need to repeat evaluations, adapt prompts and tools, retrain users and account for changes in output quality, review time and rework. For coding tools, integration with editors, repositories and team practices can also take time to rebuild.
Those costs are hard to compare without tracking work outcomes, not just bills. A company that measures only tokens may miss extra human review or errors; one that has not tested a second model may be unable to judge whether a move is worthwhile. Flexibility has a cost of its own, but building it before a contract or budget changes can make a later decision less disruptive.
As an affiliate, we earn on qualifying purchases.
Internal Adoption, Not a Customer Exit
The report concerns employees’ use and internal spending, a narrower issue than whether enterprise customers can still access Claude. It describes Meta and Microsoft moving some work toward alternatives, while Microsoft reportedly continues to use Anthropic models in features offered to customers. Those points can coexist: a company may reduce its own employees’ use while retaining a supplier’s technology in products it sells.
Both companies have commercial reasons to develop or promote other options. Meta builds its own models and coding tools; Microsoft owns GitHub Copilot and is a major backer of OpenAI. That corporate context helps explain why in-house or affiliated tools are available, but it does not prove that commercial interests were the sole reason for the reported changes. The stated drivers include cost and spending controls, and the available account does not quantify how much each factor contributed.
The reporting also describes a difference between large technology firms and typical buyers. Meta and Microsoft can draw on engineering teams and deployed substitutes. A smaller organization may have no second tool integrated into its workflows, making the same change more expensive and uncertain even if its usage bill is lower.
What the Report Does Not Establish
The available reporting does not give a complete timeline for the changes or explain how the user counts were measured. It is also unclear how much of the reported spending reduction reflects lower usage, revised forecasts, changed internal budgets or other accounting decisions. The projected Microsoft figure should not be read as money already spent or savings already realized.
Neither company is reported to have said that Claude underperformed. The reporting does not provide comparable tests of Claude, MetaCode, Muse Code, GitHub Copilot or OpenAI models on the companies’ internal tasks. It also does not quantify the engineering, training or productivity costs of the reported changes. Whether the alternatives deliver equivalent results is not answered by usage figures alone.
The status of Claude for particular teams and internal tasks is not fully described. The report does not say that either company has ended access across its workforce, and the account says Microsoft continues to use Anthropic technology in customer-facing Copilot features. The extent and terms of that use remain unspecified.
Track Usage, Costs and Results
The next useful evidence would be more detail from Meta and Microsoft on how internal use, budgets and tool assignments have changed, and whether the companies have measured output quality and productivity across alternatives. Until those details are available, the reported figures establish a shift in internal direction, not the performance of one model against another.
For businesses weighing a similar move, the practical next step is to test alternatives on representative tasks before shifting substantial work. That means keeping evaluation results, prompts and tool configurations usable across vendors, and measuring accepted work alongside token spending. A small, real-world test can reveal whether apparent savings persist once review and rework are counted.
Any further reporting should also distinguish internal adoption from customer-facing use, and projections from actual spending. Those distinctions will show whether this is a lasting change in deployment or a budget adjustment whose scope is still evolving.
Key Questions
Have Meta and Microsoft stopped using Claude?
No such company-wide end is established. The report describes reduced or redirected internal use. It says Microsoft continues to use Anthropic models for some customer-facing Copilot features.
Why are the companies reportedly reducing internal use?
The reported drivers include rising token costs, tighter spending controls and a preference for tools each company owns or backs. The report does not say either company cited poorer Claude performance.
Does the Microsoft spending figure mean it saved more than $300 million?
Not necessarily. The report describes a reduction of more than a third to a projected internal spending figure of over $1 billion a year. It does not establish realized savings or the final amount spent.
What can make switching AI models expensive?
Companies may need to repeat evaluations, adapt prompts and integrations, retrain staff and account for changes in review, rework and output quality. Those costs are not captured by token prices alone.
What remains unknown about the reported shift?
The reporting does not provide side-by-side performance results, a full timeline, or a complete accounting of actual spending and switching costs. It also does not specify how widely internal access to Claude has changed.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
