As the cost of deploying artificial intelligence continues to rise, tech giants are moving away from their previous blind enthusiasm for AI and seeking more rational spending controls. Ride-hailing giant Uber has recently introduced a strict internal regulation aimed at curbing the increasingly high costs of AI usage.

According to Uber's latest internal limit policy, each employee is strictly limited to a monthly AI usage cost of $1,500 for each intelligent code tool, including Claude Code from Anthropic and mainstream tools like Cursor. To support the implementation of this new policy, Uber has also launched an internal data dashboard, allowing employees to check their quota usage in real time. The company clearly states that only under special business scenarios and after multiple layers of approval can this limit be exceeded.

This emergency measure has attracted widespread attention in the industry, but it is not without precedent. Previously, Uber was an aggressive advocate for AI technology. Senior leaders not only encouraged employees to "use AI as much as possible" within the company, but also established AI usage rankings across departments. However, this almost indiscriminate promotion strategy soon faced financial reality. Uber's Chief Technology Officer (CTO) publicly revealed in April this year that due to the previous lax policy, the ride-hailing giant had exhausted its entire annual AI budget in just four months.

In addition to the pressure of overspending, top management has also begun to doubt the actual commercial benefits AI can bring. Uber's Chief Operating Officer (COO), Andrew McDonald, recently admitted on a podcast that it is currently difficult to clearly distinguish the direct cause-and-effect relationship between frequent AI use and actual product development. In other words, whether the high computing power investment has truly translated into productivity remains questionable within the company.

Uber's sudden tightening of AI spending reflects a common pain point in the tech industry: after years of heavy investment in AI, companies now face the soul-searching question of where the return on investment (ROI) really is. At this stage, AI's returns are still mostly theoretical expectations. When there is no tangible financial return for a long time, many companies' patience is rapidly running out.