🔍 Read the full analysis: Why Meta And Microsoft Pulled Back From Claude—and What A Switch Costs on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
The Information reported on Oct. 5 that Meta reduced internal use of Claude Code and Microsoft cut a projection for internal Anthropic spending, while both directed employees toward alternatives. The reported reasons include cost controls and available substitutes—not a stated finding that Claude performed worse. The shift shows why switching models can be easier for companies with their own tools than for typical buyers.
Meta and Microsoft have reportedly pulled back some internal use of Anthropic’s AI tools, directing employees toward alternatives including Meta’s own coding products, GitHub Copilot and OpenAI models. The Information reported the changes on Oct. 5, attributing them to cost controls and in-house options; neither company is reported to have said Claude performed worse.
Meta reportedly reduced the number of employees using Claude Code from about 60,000 earlier this year to about 30,000. The company has been steering staff toward its own 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 employee use, not customer adoption.
Microsoft had reportedly projected more than $1 billion a year in internal spending on Anthropic technology, including Claude Code, Claude models in Copilot and Claude Mythos. The report says it has since cut that projection by more than a third and is steering employees toward GitHub Copilot and OpenAI models. The source material also says Microsoft continues to use Anthropic models in customer-facing Copilot features and that customer spending on Claude through its platforms is growing.
The reported changes do not amount to a complete withdrawal of Claude access. The account describes companies redirecting some employee workloads, with rising token costs, spending controls and the availability of tools they own or back as stated drivers. Details about the timing and scope of the internal changes remain limited.
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 Requires More Than a New Model
For companies buying AI, the report highlights a gap between being able to change providers and being able to do so cheaply. Meta and Microsoft have alternatives already in use; most organizations do not have comparable in-house coding systems or the engineering resources to build them. A buyer may face costs that do not appear in a model’s price: testing workflows again, revising prompts and software integrations, and giving employees time to adapt.
There can also be costs in output quality and oversight. If a replacement model performs less well on a company’s particular tasks, the effects may appear as extra review, rework or mistakes, rather than an obvious system failure. A lower token bill does not by itself show that a switch saves money overall. Buyers need to compare the cost of producing work that is accepted, including human review.
The reported Microsoft projection gives a sense of the scale involved, but it is not a confirmed saving: the source says the forecast was cut by more than a third, not that the company has already saved that amount. For smaller buyers, the same engineering and evaluation work may be harder to justify against lower total spending.
AI coding assistant GitHub Copilot
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Internal Tools Versus Customer Products
The reporting concerns employees’ internal use of AI products. That distinction matters because companies can change what their staff use without removing a model from customer-facing services or ending customer access. The source material says Microsoft continues to spend on Anthropic models for some Copilot features, while customer use of Claude through Microsoft’s platforms is reported to be growing.
Meta develops its own models and coding tools; Microsoft owns GitHub Copilot and is a major backer of OpenAI. Their ability to steer staff to alternatives reflects those existing products and investments. The report therefore supports a narrower conclusion than a verdict on Claude’s quality: two large buyers with substitutes reportedly redirected some internal work. It does not establish that other customers are leaving Anthropic or that the tools are interchangeable for every task.
AI model training tools for enterprise
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Scope and Savings Still Unclear
The available account does not specify when each change took effect, how the usage figures were measured, or whether the reported reductions are temporary or part of a longer-term shift. Meta and Microsoft are not quoted directly in the supplied material, and it does not include company explanations beyond the drivers attributed to the reporting.
It is also unclear how the companies compared performance across products, what share of internal work remains on Claude, or whether a move has improved total costs after engineering, review and rework are counted. The reported cut to Microsoft’s spending projection does not prove a matching reduction in actual expenditure. No evidence in the material establishes that Claude underperformed on the companies’ tasks.
As an affiliate, we earn on qualifying purchases.
Watch Actual Usage and Costs
The next useful indicators are whether the reported internal shifts persist, how Meta and Microsoft distribute work among their alternatives, and whether their actual spending changes in line with the projections. Further company statements or reporting could clarify the timeline, the number of employees affected and whether customer-facing use changes.
For other organizations, the practical next step is to test more than one model on representative work before a supplier or price change forces a decision. Buyers can record task-level quality and human review effort, keep prompts and integrations portable, and run a secondary provider on a limited share of real work. That does not remove switching costs, but it can reveal them before a large migration is at stake.
As an affiliate, we earn on qualifying purchases.
Key Questions
Did Meta and Microsoft stop using Claude?
No full departure is reported. The account describes reduced or redirected internal use. It also says Microsoft continues to use Anthropic models in some customer-facing Copilot features.
Why did the companies reportedly reduce internal use?
The reported reasons include rising token costs, tighter spending controls and available alternatives. The source material does not report either company saying Claude performed worse.
How much did Meta’s Claude Code use reportedly fall?
The report puts internal use at about 60,000 employees earlier this year and about 30,000 later. The figures are reported estimates, and the account does not provide a detailed measurement method.
Does Microsoft’s spending projection show actual savings?
No. The source describes a projected annual spend of more than $1 billion that was cut by more than a third. A revised projection is not the same as confirmed savings or a tally of actual spending.
What makes switching AI models costly for other companies?
Organizations may need to retest workflows, adjust prompts and integrations, retrain users and check quality. Changes in cached context and additional review or rework can also affect the total cost.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
