The recent wave of layoffs in the tech industry might give the impression that the transition from human labor to artificial intelligence (AI) is already underway.
Last week, Meta disclosed in a memo its intention to reduce its workforce by 10%, resulting in the loss of around 8,000 jobs. Additionally, the company plans to abandon hiring for 6,000 currently open positions. This strategy aims to “enhance operational efficiency and balance the other investments we are making,” the memo stated. Similarly, Microsoft has extended an unprecedented voluntary buyout offer to thousands of its staff.
Yet, insights from tech leaders suggest that AI isn’t yet a cost-effective substitute for human workers; in fact, it may currently be more expensive.
“For my team, the costs associated with computing far exceed those of our employees,” stated Bryan Catanzaro, vice president of applied deep learning at Nvidia, in a recent interview with Axios.
An MIT study from 2024 supports Catanzaro’s perspective. Researchers found that AI models are economically viable for only 23% of roles where vision is a key component, indicating that in the remaining 77% of cases, human labor remains more cost-effective.
Moreover, there have been instances where AI has failed, such as one engineer reporting that an AI agent destroyed his database and network due to what he called “overuse.”
Despite the lack of clear evidence demonstrating AI’s impact on productivity and a report from the Yale Budget Lab showing no substantial data indicating that AI is displacing jobs, major technology firms continue to invest heavily in AI. They have announced a total of $740 billion in capital expenditures this year alone, according to Morgan Stanley, representing a 69% increase from 2025. This surge in spending has driven some companies to reassess their overall budgets.
“I’m starting over because the budget I initially anticipated has already been surpassed,” Uber’s Chief Technology Officer Praveen Neppalli Naga remarked to The Information recently, as the rideshare company shifts focus to AI coding tools like Anthropic’s Claude Code.
This increase in AI spending has coincided with a rise in layoffs within the tech sector. Data from Layoffs.fyi indicates that there have been over 92,000 job cuts in the tech industry in 2026, spanning nearly 100 companies, surpassing the previous year’s total of approximately 120,000 layoffs.
This sharp contrast between ongoing AI investments and workforce reductions, despite the current cost-effectiveness of human labor, highlights a significant imbalance in AI economics, according to Keith Lee, an AI and finance professor at the Gordon School of Business at the Swiss Institute of Artificial Intelligence.
“There’s a short-term mismatch occurring,” Lee shared with Fortune.
The AI-Labor Cost Dilemma
Lee explained that the expenses associated with AI are currently less efficient compared to human labor, primarily due to rising hardware and energy costs. By 2023, AI expenditures could reach $5.2 trillion, with $1.6 trillion attributed to data center spending and $3.3 trillion for IT equipment, as reported by McKinsey. If current trends continue, this could balloon to $7.9 trillion by 2030. Additionally, the costs of AI software have escalated by 20% to 37% in the past year, as noted by the spending management firm Tropic.
Companies specializing in AI may also face financial challenges due to their flat subscription models, as these fixed fees often fail to cover the operational costs associated with heavy AI utilization.
“Consequently, some firms are beginning to view AI not merely as a clear cost-cutting solution, but rather as a supportive tool—at least until the cost structure stabilizes,” Lee remarked.
Although AI may currently be more costly than human labor, signs indicating a potential shift towards its economic viability may emerge. Lee noted that the costs of AI usage will eventually decrease significantly, with operations related to inference for large language models expected to drop by more than 90% over the next four years, according to a recent report from analyst firm Gartner. Improvements in AI infrastructure, model designs, and hardware supply are also anticipated. Additionally, AI companies may adjust their pricing strategies, shifting from flat-rate subscriptions to a usage-based model.
However, the future economic viability of AI will also hinge on the technology proving its reliability. It must demonstrate fewer errors and require less human oversight while effectively integrating into existing company operations. Federal Reserve data shows that about 18% of companies had adopted AI tools by the end of 2025, reflecting a 68% increase in adoption since September 2025.
“It’s not solely about AI becoming less expensive than human workers,” Lee emphasized. “It’s also about achieving both cost-effectiveness and predictability at scale.”
