Companies Cut AI Spending as Big Tech Keeps Investing Billions in 2026

October 19, 2026 · 6 min read

Some of the biggest corporate AI users are tightening budgets, capping usage, and dropping tools that cost too much. Reports published on October 5 said Microsoft cut its projected internal spending on Anthropic technology by more than one-third, while fewer employees at Meta used Claude Code. At the same time, the largest tech companies are still planning record spending on AI data centres. What’s shrinking is careless, uncapped AI use inside companies, not AI investment as a whole.

What is the latest News on Companies Cutting AI Spending?

The most recent news, from this week and last, centres on Microsoft and Meta. Per the October 5 reports, Microsoft’s internal spending projection had been at least $1 billion earlier in 2026, then was reportedly cut by more than a third. Meta’s Claude Code user count reportedly fell from about 60,000 to about 30,000, partly because of layoffs and partly because of a push toward its own coding tools.

None of this is as clean as it sounds. The accounts didn’t give exact dates for either change, and the reports concern internal use only. Customer spending on Anthropic models through Microsoft’s enterprise platforms was reportedly growing. These are reports, not company announcements, so treat the specifics with some care.

Why are Companies Pulling Back on AI?

The short answer is that the bills got bigger than anyone planned. A lot of companies spent two years telling employees to experiment freely. Then the invoices arrived.

Uber is the clearest example. According to HR Reporter, Uber used up its entire planned 2026 AI coding budget within the first four months of the year because of soaring token consumption, then capped monthly spending at $1,500 per tool. A token is the basic billing unit AI companies use, roughly a chunk of text processed.

Meta’s story was messier. An internal leaderboard tracked 85,000 employees burning through more than 60 trillion tokens in 30 days before it was shut down over waste. Meta’s CTO Andrew Bosworth put the new mood bluntly: “Nobody should be using AI tools just for the sake of using them.”

What does the Spending Data Actually Show?

The best public window comes from Ramp, which tracks AI purchases across more than 70,000 U.S. businesses. The picture is mixed.

On the cautious side, Ramp’s September index found that median AI spending among the top 1% of spenders fell nearly 10% in a single month, from roughly $7,976 per employee in July to $7,205 in August.

On the other side, adoption was still rising earlier in the summer. Paid AI adoption among Ramp’s business customers hit 55.7% in July, up from just over 50% in March.

There’s one more detail worth knowing before you call this a retreat. Ramp’s economist pointed out that prices are falling. OpenAI and Anthropic have both announced price cuts, and the effective price per million tokens is down 41% to $0.68 from a March peak of $1.15. So some of the per-employee decline could be cheaper AI rather than less AI.

Keep the sample in mind as well. Ramp says its data skews toward the technology industry and excludes large enterprises whose AI spending runs through other providers. It’s a useful signal, not the whole market.

Is Big Tech Actually Spending Less on AI?

No. This is the part that gets lost. The same companies trimming internal tool budgets are pouring money into infrastructure.

According to a Financial Times tally, Amazon, Microsoft, Alphabet and Meta collectively plan roughly $725 billion in 2026 capital spending, up 77% from last year’s record of about $410 billion. That was after April earnings. Company by company, Microsoft expects about $190 billion, Alphabet $180 billion to $190 billion, Amazon close to $200 billion, and Meta $125 billion to $145 billion.

Those are guidance figures, not booked spending. The same reporting notes that they don’t show how much is already contracted versus optional. Analysts have also warned that if AI revenue doesn’t grow fast enough, this cycle could start to look like an overbuild.

So both things are true. Companies are getting stricter about buying and using AI tools, and the companies selling the infrastructure behind them are still building at enormous scale.

How Does This Connect to AI Layoffs?

It’s a related but separate story, and the two get mixed up. Cutting jobs and cutting AI budgets can happen at the same company, sometimes for the same reason: pressure to show returns.

On layoffs, the numbers are large. Challenger, Gray & Christmas data cited by Entrepreneur shows that by the end of September, AI had been cited in more than 120,000 announced U.S. job cuts in 2026, about 21% of the total, making it the leading stated reason this year. High-profile examples include Amazon’s nearly 16,000 cuts, Meta’s plan to eliminate about 8,000 positions, and Block’s more than 4,000.

A caution applies here too. Some critics say companies are “AI washing,” meaning blaming layoffs on AI productivity gains when the real motive may be cost-cutting and profitability. Quietly hiring less is also a big factor. New York Fed surveys found that service firms using AI were almost four times as likely to hire fewer workers as to lay any off.

Does this Mean the AI Boom is Ending?

Probably not, and the evidence points toward a correction in habits rather than a collapse. One analysis of the Fable 5 pricing question made a useful distinction: the resistance is to one premium model, not to AI spending as a category. Buyers are getting pickier about which models they pay for.

Cost discipline is what a maturing technology looks like. Companies that treated AI like free electricity are learning to meter it, set caps, and ask what each dollar returns. The ones that do well will likely be those that match cheaper models to simpler tasks and reserve premium ones for work that justifies them.

What should Businesses do Now?

Based on what the reporting shows, a few practical moves stand out:

Set usage caps early. Uber’s per-tool limit came after the budget was already gone. It’s easier to cap first.
Track cost per outcome, not cost per seat. Token usage tells you nothing about value.
Route work by difficulty. Don’t send routine tasks to the most expensive model.
Centralize ownership. CFOs are doing this for a reason: scattered team-level subscriptions hide the real total.
Review vendor pricing often. With token prices falling 41% in about six months, last quarter’s contract may already be out of date.

FAQ

Are companies reducing AI spending in 2026?

Some are, especially on internal tools and high-cost usage. Infrastructure spending by the largest tech companies is still rising sharply.

Which companies have cut AI spending?

Reports name Uber (budget exhausted by April, then capped), Microsoft (reduced internal Anthropic spending), and Meta (fewer Claude Code users). Microsoft’s and Meta’s changes are reported by media, not confirmed in detail by the companies.

Is AI the top reason for layoffs in 2026?

According to Challenger data through September, yes. It was cited in about 21% of announced U.S. job cuts, though some analysts question how much of that is genuinely AI-driven.

Source:

Topics:

Reviewed by editorial staff before publication.
Fact-checking and source verification applied.
Updated regularly for accuracy and clarity.
Aligned with newsroom ethics and publishing standards.

About The Author

Jordan Drew Senior Editor

Jordan Drew is a Technology and Economy Correspondent at New York Editor, covering technology, business, finance, markets, digital innovation, and major economic developments. Jordan focuses on explaining how emerging technologies and changing economic trends are influencing businesses, consumers, and the wider world. With an interest in both innovation and the forces shaping modern economies, Jordan produces clear, informative, and well-researched stories that make complex topics easier to understand. Their coverage includes artificial intelligence, consumer technology, startups, digital markets, business trends, economic policies, and the growing relationship between technology and the global economy. Jordan believes good technology and economy journalism should go beyond the headlines by giving readers useful context and helping them understand what developments mean in real life. "Technology is changing more than the way we work and communicate. My goal is to help readers understand the ideas, innovations, and economic trends shaping the world around them."

On This Page