The high cost of artificial intelligence has forced ride-hailing giant Uber to restrict its staff from using advanced AI tools freely. The firm has introduced a strict monthly spending limit of 1,500 dollars (around Rs 1.43 Lakh) per employee for specialized AI coding assistants.
The emergency measure follows a frank admission from Uber's Chief Technology Officer, who revealed the company had managed to spend its entire annual AI budget in just four months.
The policy change highlights a growing anxiety within the technology sector. While major corporations have rushed to adopt AI tools, many are now questioning whether the staggering financial investment is delivering any real value to their business.
Competitive pressure led to spiraling costs
According to a report by TechCrunch, Uber initially encouraged its employees to use AI tools as much as possible. The company went so far as to rank staff members on internal leaderboards to spark competition and drive up usage.
However, this aggressive push backfired financially. Software engineers heavily relied on advanced platforms such as Anthropic's Claude Code and Cursor to write and debug software. Because these systems charge based on the complexity and volume of data processed, the continuous usage caused costs to spiral out of control.
To stop the financial bleed, Uber has launched a digital dashboard to track how much money each worker spends on AI every month. Staff members will now need special permission if they want to exceed the 1,500-dollar threshold.
Is AI actually worth the money?
The budget crisis has sparked a wider debate inside Uber about the actual usefulness of artificial intelligence. Andrew Macdonald, the company's Chief Operating Officer, recently admitted that it is incredibly difficult to connect the heavy use of AI with any visible improvements for everyday consumers.
Uber is not alone in facing this dilemma. For the past two years, tech companies have told workers that mastering AI is essential for survival. This psychological pressure created a workplace culture where staff felt compelled to use the technology for every task. Now, companies are realizing that the daily operational costs of running these advanced systems are often higher than the human productivity gains they provide.
For now, the tech industry's massive bet on artificial intelligence remains an expensive experiment, and Uber's sudden pullback suggests that the era of unlimited corporate AI spending is officially over.
Why agentic AI tools cost so much to run
Agentic tools process massive volumes of data because they do not just answer questions; they autonomously execute complex, multi-step workflows. Every action-such as reading a file, running a test, or fixing an error-requires a separate call to a premium AI model like Claude 3.5 Sonnet. Because these systems charge per token, and must repeatedly resend the entire codebase and conversation history at every step of the loop, a single user request can trigger exponential, spiraling costs.
