Uber Limits AI Coding Tool Spend to $1,500 Per Employee a Month: Report

Uber Technologies has introduced spending caps on AI-powered coding tools used by employees after the company exhausted its AI coding budget earlier this year, according to a Bloomberg report.

The ride-hailing company is limiting staff to $1,500 per month in token spending per AI coding tool, an Uber spokesperson told Bloomberg. 

The limits apply separately to each tool, meaning usage on one platform does not reduce an employee's allowance on another. 

The restrictions were introduced over recent months and apply to agentic coding products such as Anthropic’s Claude Code and Cursor.

Unlike conventional chatbot deployments, coding agents often consume large volumes of tokens while carrying out multi-step tasks, increasing infrastructure and software expenses even as the underlying model. 

The spending limits follow comments from Uber executives earlier this year that the company had already exhausted its planned 2026 AI coding tools budget within the first four months of the year.

Uber CTO Praveen Neppalli Naga has said surging adoption of tools such as Claude Code had pushed spending well beyond internal forecasts, while the company expanded AI-assisted software development across its engineering organisation. 

Reports at the time indicated that a large majority of Uber engineers were regularly using AI coding tools, driving a sharp increase in token consumption and software costs.

Uber's move comes amid a broader industry debate over “tokenmaxxing”—the practice of maximising AI model usage on the belief that higher token consumption leads to greater productivity. 

Uber COO Andrew Macdonald recently questioned whether rising AI usage could be directly linked to more useful consumer-facing products, arguing that companies will increasingly need to justify AI spending against measurable business outcomes. 

Similar discussions have emerged across large technology companies as agentic coding systems generate rapidly growing infrastructure and API costs.

The debate has spread beyond Uber. Microsoft has reportedly reduced internal Claude Code deployments amid rising AI costs, while Google executives have acknowledged growing enterprise scrutiny of AI spending and returns. 

According to a report from Axios, Salesforce has emphasised the ROI on AI usage and efficiency rather than raw model consumption. 

The company has even criticised ‘tokenmaxxing’ the idea that more AI usage automatically means more value, and introduced internal metrics focused on business outcomes instead of token volume.

Across the industry, companies are increasingly evaluating whether higher token consumption and broader AI adoption are translating into measurable productivity gains and business outcomes.