The belief that AI would quickly reduce company expenses is now facing serious doubts. Many tech firms expected AI tools to replace repetitive work and improve employee productivity. However, companies are now discovering that AI systems require massive spending while delivering limited results in day-to-day operations. Microsoft recently asked thousands of engineers to stop using Claude Code and shift to its internal AI tools by June 30. Reports suggest the companies grew concerned about the rising token costs linked to Claude usage. Managers reportedly noticed that employees were consuming expensive AI resources without showing significant productivity improvements.
Tech firms struggle to justify soaring AI expenses amid weak productivity results.
Uber is facing a similar challenge with its AI spending. The company introduced Claude Code for nearly 5,000 engineers earlier this year. Within just five months, Uber exhausted its complete AI budget planned for 2026. Company executives admitted that they are struggling to connect AI spending with meaningful business outcomes.
Industry experts say token-based pricing has become one of the biggest concerns for companies using AI at scale. AI firms like OpenAI, Anthropic, and Google now charge based on token usage instead of flat subscription plans. As employees use AI tools more frequently, businesses are seeing operational costs rise sharply.
Several tech leaders believe AI tools can sometimes cost more than hiring human workers. Nvidia executive Bryan Catanzaro recently stated that compute costs for AI systems are now higher than employee salaries in some cases. This comparison has pushed companies to rethink how aggressively they should expand AI adoption.
Despite the growing expenses, experts do not believe the AI boom is ending. Instead, companies are entering a new phase where they must carefully evaluate the return on investment. Business leaders now want clear evidence that AI tools improve productivity and generate measurable results before increasing spending further.
Also Read :India, U.S. Sign Critical Minerals Partnership
Industry insiders say companies must focus on smarter AI deployment rather than uncontrolled expansion. Experts suggest that businesses should select practical use cases and monitor token consumption closely. They believe AI can still provide long-term value, but only if organisations manage costs and usage more efficiently.