The tech industry’s aggressive push into artificial intelligence is creating a paradox that few saw coming: massive capital spending on AI infrastructure is coinciding with widespread layoffs, even as many companies admit that human labor remains cheaper than AI in most real-world applications today.
Meta’s announcement last week that it would cut roughly 10% of its workforce, about 8,000 jobs, and scrap plans to fill 6,000 open positions was framed internally as a necessary efficiency move. In the memo, the company said the reductions would help “run the company more efficiently and to allow us to offset the other investments we’re making,” a thinly veiled reference to its enormous AI outlays.
Microsoft has offered thousands of employees a voluntary buyout — the largest in the company’s history. Across the sector, Layoffs.fyi data shows more than 92,000 tech jobs have already been eliminated in 2026, a pace that is outstripping last year’s total of around 120,000 cuts.
Register for Tekedia Mini-MBA edition 20 (June 8 – Sept 5, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab.
At first glance, the numbers suggest the long-predicted shift from human workers to AI is already underway. But conversations with executives and analysts reveal a more complicated picture: AI is not yet delivering clear cost savings. In many cases, it is costing companies more than the humans it might eventually replace.
Nvidia vice president of applied deep learning Bryan Catanzaro put it plainly in a recent Axios interview. He said: “For my team, the cost of compute is far beyond the costs of the employees.”
An MIT study from 2024 reached a similar conclusion. After analyzing the technical requirements for AI to match human performance, researchers found that automation would be economically viable in only 23% of roles where vision is a primary component. In the other 77%, it was still cheaper to keep humans in the job.
According to Fortune, Keith Lee, an AI and finance professor at the Swiss Institute of Artificial Intelligence’s Gordon School of Business, described the situation as a classic short-term mismatch.
“What we’re seeing is a short-term mismatch,” Lee told Fortune.
AI companies are often losing money on flat subscription models that fail to cover the high operating costs for heavy users. As a result, some firms are starting to view AI more as a complementary tool rather than an immediate labor substitute.
The scale of the spending is staggering. The four major U.S. tech giants that reported earnings this week, Alphabet, Meta, Amazon, and Microsoft, have collectively signaled AI-related capital expenditures that are now projected to top $700 billion this year, up from around $600 billion previously. Alphabet raised its annual capex forecast by $5 billion to between $180 billion and $190 billion, with plans for another big increase in 2027. Microsoft expects $190 billion in 2026 spending, with roughly $25 billion tied to rising component costs. Meta lifted its ceiling to as much as $145 billion.
Uber chief technology officer Praveen Neppalli Naga recently told The Information that the company’s pivot to AI coding tools had blown up its budget.
“I’m back to the drawing board because the budget I thought I would need is blown away already,” he said.
According to McKinsey projections, AI expenditures could reach $5.2 trillion globally by 2030 in a base case, or as high as $7.9 trillion at an accelerated pace. AI software fees have already risen 20% to 37% over the past year, according to Tropic.
Despite the spending spree, widespread productivity gains or large-scale job displacement have not yet materialized. The Yale Budget Lab has pointed to a lack of robust data supporting the idea of AI broadly replacing workers. Federal Reserve figures show that only about 18% of companies had adopted AI tools by the end of 2025, a 68% increase since September, but adoption remains early-stage and uneven.
Lee sees a clear path toward AI becoming economically superior, but it will take time and several breakthroughs. Inference costs for large language models with 1 trillion parameters are expected to drop more than 90% over the next four years, according to Gartner. Improvements in infrastructure, model efficiency, and hardware supply will help, and pricing models are likely to shift from flat subscriptions to usage-based structures that better align costs with actual value delivered.
But viability will ultimately depend on reliability.
“It’s not just about AI becoming cheaper than humans,” Lee said. “It’s about becoming both cheaper and more predictable at scale.”
For now, companies are making a high-stakes bet on that future. Google Cloud’s 63% revenue surge in the March quarter, far above estimates, was driven primarily by AI tools for enterprises for the first time, vindicating Alphabet’s heavy investment in turning research into commercial products. CEO Sundar Pichai noted that capacity constraints limited even stronger growth, a problem echoed across the industry.
Analyst Lee Sustar of Forrester observed that Google is capturing new workloads, sometimes from companies new to the cloud or seeking to diversify.
“It is capturing new workloads for the most part — sometimes from companies new to cloud, often additional workloads from customers of other clouds who want to be less dependent on a single cloud provider or who like Google data, analytics and AI offerings,” he said.
The current wave of heavy spending and selective layoffs reflects a painful transition period. Companies are investing aggressively in AI while trimming costs elsewhere to protect margins and reassure investors. Human labor remains cheaper and more reliable for many tasks today, but the scale of the capital bets suggests executives believe the economics will eventually flip as the technology matures.
The discrepancy between soaring AI costs and continued reliance on human workers underscores that the great labor shift to AI is not happening overnight. It is a multi-year, high-risk wager on future efficiency gains that have yet to fully materialize. Currently, the most visible impact is not mass replacement of workers, but a costly arms race to build the infrastructure that might one day make AI the cheaper, more scalable option.