Image Courtesy: Getty Images
A senior Nvidia executive says that, despite widespread expectations, artificial intelligence is not yet reducing labor costs. In fact, for many teams, it is doing the opposite. Bryan Catanzaro, Nvidia’s vice president of applied deep learning, stated that the cost of running AI systems currently exceeds the cost of employing human workers.
The comment highlights a growing gap between the promise of automation and its present-day economics. AI has been widely positioned as a tool to streamline operations and reduce headcount, but real-world deployment is proving more expensive than anticipated. Catanzaro noted that compute costs, which include the infrastructure required to train and run AI models, are significantly higher than employee salaries in his own team’s case, as reported by Fortune.
Research findings support this view. A 2024 study from the Massachusetts Institute of Technology examined the financial viability of replacing human workers with AI across various roles. It found that automation made economic sense in only 23 percent of jobs that rely heavily on visual tasks. In the remaining 77 percent, human labor remained the more cost-effective option.
There are also operational risks tied to AI adoption. Instances of costly errors have surfaced, including cases where AI tools caused data loss or system failures. These risks add to the overall cost burden, especially when factoring in recovery and oversight.
Despite these challenges, companies are continuing to invest heavily in AI. Morgan Stanley estimates that tech firms have committed around $740 billion to AI-related spending this year alone, marking a sharp increase from 2025. Software costs are also rising, with AI-related fees increasing between 20 percent and 37 percent over the past year, based on data from spending management firm Tropic.
Long-term projections suggest even larger investments ahead. McKinsey estimates total AI spending could reach $5.2 trillion by 2033, including substantial allocations for data centers and hardware infrastructure. However, this level of spending is forcing some companies to reassess their budgets. Uber’s chief technology officer, Praveen Neppalli Naga, recently said that the company’s shift toward AI coding tools has already exceeded initial cost expectations.
At the same time, layoffs in the tech sector are accelerating. Data from Layoffs.fyi indicates that more than 92,000 tech workers have lost their jobs this year across nearly 100 companies. That pace is significantly faster than the previous year’s total.
Experts say this reflects a short-term mismatch between cost efficiency and strategic direction. Keith Lee, an AI and finance professor at the Swiss Institute of Artificial Intelligence’s business school, noted that companies are investing heavily in AI even when it is not yet the cheaper option. The expectation is that costs will decline as infrastructure improves and systems become more reliable.
For now, the economics remain unsettled. AI may eventually become both cheaper and more predictable, but current data suggests that, in many cases, human workers still offer better value.