Companies like Uber and Nvidia reveal that expenses for artificial intelligence are surpassing payroll costs.

n n

Companies accelerating their adoption of Artificial Intelligence are discovering that the cost of AI tools and computing power can quickly exceed employee salaries, according to executives from Nvidia and Uber.

n

As reported by The Economic Times on Friday, Bryan Catanzaro, Vice-President of Applied Deep Learning at Nvidia, told Axios that AI compute expenses have become a major burden for some teams.

n

“For my team, the cost of compute is far beyond the costs of the employees,” Catanzaro said.

n

His comments reflect growing concerns around the economics of AI automation, particularly as businesses expand their use of coding assistants, AI agents, and automated workflows.

n

Unlike consumer AI chatbots that typically operate on flat subscription plans, enterprise AI spending is often based on token usage.

n

Costs rise significantly as companies run continuous AI-driven coding, automation, and scheduled tasks.

n

At Uber, the company’s AI spending estimates have already increased sharply.

n

Uber’s Chief Technology Officer, Praveen Naga, said the company had underestimated how much AI adoption would cost.

n

“(I’ve) gone back to the drawing board because the budget (I) thought (I) would need is blown away already,” Naga said.

n

The issue is also affecting startups and smaller companies experimenting heavily with AI tools. In a LinkedIn post, Swan AI founder Amos Bar-Joseph revealed that his four-person company received a $113,000 bill from Anthropic, the company behind Claude AI.

n

That works out to roughly $28,000 per employee in a single month.

n

The rising costs have reignited debate over whether AI is truly cheaper than human labour.

n

A 2024 study by researchers at the Massachusetts Institute of Technology found that humans completed tasks more efficiently in 77% of cases cited in the report.

n

Despite the concerns, companies continue investing aggressively in automation. Uber’s Naga said AI agents are already responsible for 11% of the company’s live code updates.

n

“The vision for [me] as a CTO is to transform from software engineering to [AI] agent software engineering,” he said.

n

Meanwhile, Jensen Huang is reportedly encouraging engineers to increase their use of AI tools. According to the report, Huang wants engineers earning $500,000 annually to spend at least $250,000 a year on AI tokens.

n

Businesses are now weighing whether today’s AI expenses are temporary investments before automation becomes more efficient, or whether high token and compute costs will remain a permanent addition to operating expenses.

n

At the same time, companies that rush into AI adoption without clear strategies risk significant financial losses, as many firms are still struggling to generate meaningful returns from their AI investments.

n

Even so, layoffs linked to automation are expected to continue as companies test the long-term economics of AI-driven workforce reduction.

n n