Uber Caps AI Tool Usage After Soaring Costs

Artificial intelligence may be transforming workplaces, but the growing cost of using advanced AI systems is beginning to worry large companies. Ride-hailing giant Uber has reportedly started restricting employee access to premium AI coding tools such as Claude Code and Cursor after the company burned through its yearly AI budget in just four months.

According to a Bloomberg report, Uber has introduced a monthly spending limit of $1,500 per employee for each AI coding platform. The cap specifically targets agentic AI tools capable of independently writing, reviewing, and editing software code with minimal human intervention.

Uber reportedly said the move is designed to encourage “responsible experimentation” while helping the company manage rapidly increasing AI expenses as adoption expands across departments.

Importantly, the $1,500 cap applies separately to each AI tool. Employees using multiple platforms will receive individual budgets for every service. Staff members who require additional access can seek approval to exceed the assigned limit. Uber has also rolled out internal dashboards that allow employees to monitor their AI usage and related spending in real time.

The decision marks a sharp shift from Uber’s earlier push toward aggressive AI adoption. Earlier this year, the company strongly encouraged employees to integrate AI into daily workflows. Internal leaderboards reportedly ranked staff based on how extensively they used AI systems.

In April, Uber Chief Technology Officer Praveen Neppalli Naga disclosed that the company had already exhausted its annual AI budget within the first four months of 2026.

Uber executives have consistently highlighted AI’s growing role inside the organisation. Chief Executive Officer Dara Khosrowshahi recently stated that nearly 10 per cent of Uber’s code is now generated and submitted by AI agents. Beyond software engineering, departments including legal and marketing have also adopted AI-powered tools to streamline repetitive tasks and improve operational efficiency.

Despite the rapid adoption, Uber admits that measuring the actual business value of AI remains difficult. Speaking on the Rapid Response podcast, Chief Operating Officer Andrew Macdonald said the company has struggled to directly link increased AI usage with meaningful customer-facing improvements. Although internal productivity indicators have improved, the long-term impact of AI investments is still unclear.

Uber’s situation reflects a broader challenge facing the tech industry. As companies move from AI experimentation to large-scale deployment, many are discovering that advanced AI systems come with substantial recurring expenses. Since most enterprise AI services rely on usage-based pricing, costs increase sharply as employee adoption grows.

Microsoft is reportedly dealing with similar concerns. Recent reports suggest the company has restricted internal access to Anthropic’s Claude Code and asked employees to shift toward GitHub Copilot CLI instead.

The change mainly affects engineers working within Microsoft’s Experiences + Devices division, which oversees major products such as Windows, Microsoft 365, Outlook, Teams, and Surface devices.

Microsoft has reportedly set June 30, 2026, as the transition deadline, aligning the move with the close of its fiscal year.

While Microsoft publicly described the transition as an effort to standardise developer tools, reports indicate that rising operational costs played a significant role. Claude Code reportedly gained massive popularity among engineers after its expansion in late 2025, but the heavy token usage associated with complex AI workflows eventually pushed costs sharply higher.

The developments at both Uber and Microsoft underline a growing reality for the technology industry: while AI promises massive productivity gains, sustaining large-scale AI adoption may prove far more expensive than many companies initially expected.