For the past two years, Wall Street has embraced a simple AI investment thesis: the more companies adopt artificial intelligence, the more productive they become. That narrative has helped propel Nvidia Corp. (NASDAQ:NVDA) to the center of the market’s AI boom.
But some early signs suggest there may be a catch.
AI Adoption Doesn’t Automatically Mean Lower Costs
One of the biggest promises surrounding generative AI has been its ability to help companies do more with fewer people. Investors have rewarded firms that touted AI-driven productivity gains, while tech leaders have repeatedly pointed to automation as a way to boost efficiency.
Yet some of the companies deploying AI at scale are reporting a different challenge: rapidly rising AI bills.
Speaking to CNN, Axios Senior AI Reporter Madison Mills said she first heard concerns from a Nvidia executive.
“I spoke with a VP at Nvidia who first flagged this to me. He said, ‘Oh yeah, for months our costs for my team have been more for AI than humans,’” Mills said.
The comment echoes a similar observation made publicly by Nvidia Vice President of Applied Deep Learning Bryan Catanzaro, who recently told Axios that for his team, compute costs have grown beyond employee costs.
Uber’s AI Budget Was Gone By April
Uber Technologies Inc. (NYSE:UBER) may offer one of the clearest examples of the emerging trend.
According to reports from The Information, Uber rolled out Anthropic‘s Claude Code to engineers late last year and quickly saw adoption surge. By March, roughly 84% of the company’s engineers were using the tool, with AI contributing to a large share of committed code.
The popularity came with a price tag.
Uber CTO Praveen Neppalli Naga reportedly said the company’s planned AI budget for the year had effectively been exhausted by April, with heavy users generating monthly AI expenses that reached into the thousands of dollars.
The experience highlights a new reality of AI economics: unlike traditional software subscriptions, many AI tools charge based on usage. Every prompt, code review and agent-driven workflow consumes tokens and incurs additional costs.
Microsoft’s Claude Retreat Raises Questions
Microsoft Corp. (NASDAQ:MSFT), one of AI’s biggest backers and a major investor in Anthropic rival OpenAI, is also reportedly reducing internal use of Anthropic’s Claude Code in favor of its own GitHub Copilot tools.
While Microsoft has framed the move around standardization and tooling strategy, reports have suggested rising AI software costs may have played a role.
The irony is hard to miss. The same industry that has spent years promoting AI as a cost-saving technology is increasingly confronting the costs that come with widespread adoption.
A Bullish Signal For Nvidia — And A Risk For Everyone Else
For Nvidia, exploding AI usage remains overwhelmingly positive. Every additional AI query, agent workflow and coding session ultimately drives demand for the chips and infrastructure powering the technology.
For enterprise customers, however, the equation may be more complicated.
The more employees rely on AI, the more compute they consume. And while the productivity gains may be real, so are the bills.
That creates a question investors may soon start asking alongside revenue growth and AI adoption metrics:
If AI is making workers more productive, will it also make companies more profitable?
As AI usage expands across corporate America, the answer may depend less on how much AI gets used — and more on who ends up paying for it.
Photo: Blossom Stock Studio / Shutterstock
Recent Comments