Recent layoffs in the technology sector appear to indicate that a large-scale labor migration from initially human workers to AI may already be occurring.
In a memo earlier this year, Meta announced plans to lay off about 8,000 people, or 10% of its workforce, and scrap plans to fill 6,000 open positions. This is part of an effort to “run the company more efficiently and help offset other investments we are making,” the memo said. Microsoft has offered thousands of its employees in the largest voluntary buyout the company has ever offered.
However, other technology headers suggest that AI is not currently saving companies labor costs. In fact, it costs more than the humans we currently employ.
“For my team, the cost of computing far exceeds the cost of employees,” said Brian Catanzaro, vice president of applied deep learning at Nvidia. Axios In April.
A 2024 MIT study confirms the experiences of these executives. Analyzing the technical requirements for AI models needed to perform jobs at a human level, researchers found that AI automation is economically viable for only 23% of roles where vision is a major part of the job. The other 77% of the time, it was cheaper for humans to continue doing their jobs.
In other cases, AI has proven fallible, with one engineer saying AI agents destroyed databases and networks as a result of what he called “overuse.”
Why are tech companies pouring money into AI despite financial pressure?
Despite a lack of clear evidence that AI improves productivity and no widespread data to support the idea that AI will displace jobs, according to the Yale Budget Institute, Big Tech companies continue to pour money into AI, with $740 billion in capital spending announced so far this year, a 69% increase from 2025, according to Morgan Stanley. The size of the expense has caused some companies to completely rethink their budgets.
“The budget we thought we needed has been blown and we’re back to square one,” said Praveen Nepali Naga, Uber’s chief technology officer. information Earlier this month, we noted that the ride-sharing giant was pivoting to AI coding tools like Anthropic’s Claude Code. Naga said the company used its 2026 AI coding tools budget by April as a result of encouraging adoption through a leaderboard for employees adopting AI coding tools.
Microsoft is also canceling most of its direct licenses for cloud code. The Verge was reported last month and migrated to the GitHub Copilot CLI instead. As the technology company pushed its employees to integrate AI into their workflows, the technology essentially became too popular.
This increase in spending coincides with increased layoffs in the high-tech sector. According to data from Layoffs.fyi, the technology industry has cut more than 118,000 jobs at about 100 companies so far in 2026. The pace of job cuts is already far outpacing last year’s pace, when a total of about 120,000 people were laid off.
Keith Lee, professor of AI and finance at the Gordon School of Business at the Swiss Institute for Artificial Intelligence, said the continued spending on AI and layoffs, even though human labor remains cheap, exposes a major contradiction in the economics of AI.
“What we’re seeing is a short-term mismatch,” Lee said. luck.
When will there be a balance between AI and labor costs?
According to Lee, the cost of using AI remains less efficient than human labor because hardware and energy drive up operating costs for providers. At the current pace, AI spending could reach $5.2 trillion by 2030, including $1.6 trillion in data center spending and $3.3 trillion in IT equipment spending, according to McKinsey data. Spending could accelerate to $7.9 trillion by 2030. Meanwhile, spend management firm Tropic noted in December 2025 that fees for AI software have increased by 20% to 37% over the past year.
Lee also pointed out that AI companies may also be losing money as a result of fixed subscription models, as fixed subscription fees cannot cover the operating costs of heavy AI users.
“As a result, some companies are starting to re-evaluate AI not as an obvious cost-savings replacement for labor, but as a complementary tool, at least until cost structures stabilize,” he said.
While AI may cost more than human labor today, there will be warning signs of a tipping point toward AI’s economic viability. As an example, Lee pointed out that a report last month from analytics firm Gartner showed that the cost of running inference (the way AI analyzes data) for large language models with 1 trillion parameters will plummet by more than 90% over the next four years, making the cost of using AI much lower. AI infrastructure will likely improve, and model design and hardware supply will likely follow. Lee predicted that AI companies will also likely change the way they price their tools, moving from flat-rate subscriptions to pay-as-you-go pricing.
But the future of AI’s economic viability also depends on whether the technology can prove its worth. Lee says it will reduce hallucinations, reduce the need for human oversight, integrate effectively into a company’s infrastructure, and need to prove trustworthiness. Approximately 18% of companies had adopted AI tools at the end of 2025, according to data from the Federal Reserve Board, with adoption rates increasing by 68% since September 2025.
“It’s not just that AI costs less than humans,” Lee said. “It’s about being cheaper and more predictable at scale.”
A version of this article was published on Fortune.com on April 28, 2026.
AI cost details:
- ‘Big Tech is hopeless’: Amazon engineers slam tech giant for $200 billion in data center spending after cutting 30,000 employees
- AI productivity gains are real, but so is bad management. “Leaders have a really hard time articulating what their vision and strategy is.”
- No matter how you look at it, U.S. companies are winning with AI adoption, but a series of high-profile fracas shows they’re taking a hit on the cost front.
