Key Takeaways
- Nvidia vice president Bryan Catanzaro says that for his team, AI compute now costs more than the employees using it, making AI more expensive than human labor.
- A 2024 MIT study finds AI automation is economically viable in only about 23% of jobs, with humans still cheaper in the remaining 77%.
- Despite unclear productivity gains and high costs, big tech companies have committed around $740 billion to AI-related expenses this year, a 69% jump from 2025.
A key Nvidia executive says that AI isn’t reducing labor costs — right now, it’s actually more expensive than the human workers that companies already have.
“For my team, the cost of compute is far beyond the costs of the employees,” Bryan Catanzaro, vice president of applied deep learning at Nvidia, recently told Axios.
A 2024 MIT study supports this view. Researchers looked at what it would take for AI systems to match human performance across different jobs and found that automation made financial sense in just 23% of roles that rely heavily on visual tasks. In the other 77% of cases, keeping human workers was still the more cost-effective option.
There are also examples of AI making costly errors. In one case, an engineer said an AI tool wiped out his database and network.
Companies are investing in AI
Despite the drawbacks of AI, big tech companies are still investing heavily in it. According to Morgan Stanley, tech firms have already committed about $740 billion to AI-related spending this year, a 69% increase from 2025. Over just the past year, fees for AI software have also gone up sharply, increasing by 20% to 37%, according to spending management company Tropic.
AI spending is rising quickly. Based on McKinsey estimates, it could reach $5.2 trillion by 2033, including about $1.6 trillion for data centers and $3.3 trillion for IT hardware.
The scale of that investment is so large that some companies are now reconsidering how they allocate their budgets. For example, Uber’s chief technology officer, Praveen Neppalli Naga, told The Information earlier this month that the company’s shift toward AI coding tools is driving up costs. “I’m back to the drawing board because the budget I thought I would need is blown away already,” he said.
Tech layoffs are on the rise
As companies increase AI spending, layoffs across the tech industry have been rising. Data from layoff tracker Layoffs.fyi shows that more than 92,000 tech workers have already lost their jobs this year, spanning nearly 100 companies. That pace is much faster than last year, when roughly 120,000 layoffs occurred over the entire year.
Keith Lee, an AI and finance professor at the Swiss Institute of Artificial Intelligence’s Gordon School of Business, told Fortune that companies are spending huge amounts on AI, even though human workers are currently cheaper for many tasks. There’s a gap between what makes financial sense on paper and what companies are actually doing. “What we’re seeing is a short-term mismatch,” Lee told the outlet.
AI may be more expensive than human workers right now, but that could change. Lee says the economics will shift as the cost of running AI models drops and infrastructure improves. He added that AI will only become truly cost-effective if it proves reliable and needs less human supervision.
“It’s not just about AI becoming cheaper than humans,” Lee told Fortune. “It’s about becoming both cheaper and more predictable at scale.”
Key Takeaways
- Nvidia vice president Bryan Catanzaro says that for his team, AI compute now costs more than the employees using it, making AI more expensive than human labor.
- A 2024 MIT study finds AI automation is economically viable in only about 23% of jobs, with humans still cheaper in the remaining 77%.
- Despite unclear productivity gains and high costs, big tech companies have committed around $740 billion to AI-related expenses this year, a 69% jump from 2025.
A key Nvidia executive says that AI isn’t reducing labor costs — right now, it’s actually more expensive than the human workers that companies already have.
“For my team, the cost of compute is far beyond the costs of the employees,” Bryan Catanzaro, vice president of applied deep learning at Nvidia, recently told Axios.
A 2024 MIT study supports this view. Researchers looked at what it would take for AI systems to match human performance across different jobs and found that automation made financial sense in just 23% of roles that rely heavily on visual tasks. In the other 77% of cases, keeping human workers was still the more cost-effective option.
