Uber reveals an unexpected problem behind the AI boom | Official Website of Louis Velazquez, entrepreneur, finance guy and tech innovator

Every company in America is being told to embrace AI. Very few are being honest about what will happen when the bill arrives.

Uber (UBER) just became one of the first major technology companies to publicly acknowledge a problem that is quietly spreading across corporate America: AI tools are genuinely useful, expensive in ways that budgets weren’t built for, and remarkably difficult to connect to measurable business outcomes.

The rideshare Goliath blew through its entire 2026 AI budget in four months, according to Bloomberg reporting. It has since implemented $ 1,500-per-tool-per-month spending caps for all employees on agentic coding software, such as Anthropic’s Claude Code and Cursor.

Approximately 95% of Uber’s 5,000 engineers now actively use AI-assisted coding tools monthly, according to Forbes. In May, a Business Insider report showed that Chief Executive Dara Khosrowshahi said roughly 10% of the company’s code is now submitted and built by AI agents.

And yet, asked whether all of this is translating into better products for customers, Chief Operating Officer Andrew Macdonald offered an unusually candid answer.

“It’s very hard to draw a line between one of those stats and ‘OK, now we’re actually producing like 25% more useful consumer features,'” Macdonald said, according to Business Insider.

That gap between measurable individual productivity and uncertain organizational outcomes is the unexpected problem the AI boom is producing. Uber just said it out loud.

Also Read: History of Uber: Timeline and Facts

How Uber burned through its 2026 AI budget in four months

The budget depletion wasn’t the result of reckless spending. It was the result of adoption that far exceeded what anyone predicted when 2026 budgets were set in 2025.

The $1,500 monthly cap Uber has now implemented applies per tool, per employee. My review of the math reveals what this means in practice.

If an engineer uses two tools — say, Claude Code and Cursor — the annual cap is approximately $36,000 per engineer. Median compensation for Uber software engineers in the U.S. runs approximately $330,000 annually, according to Levels.fyi Uber Compensation report.

More Automotive:

That means the AI spending cap per engineer is roughly 11% of median total compensation — a significant line item that wasn’t in any budget forecast two years ago.

Uber has provided every employee with a usage dashboard to track token consumption across tools. Individuals can apply to exceed their caps.

“We think this is all a pretty straightforward way to responsibly encourage agentic AI adoption and experimentation at scale across the company,” an Uber spokesperson told Bloomberg.

The company is also moderating its overall hiring pace relative to plans entering the year — a direct consequence of AI-driven productivity gains that reduce headcount requirements.

Hyperscalers are investing heavily in AI infrastructure, with spending expected between $700 billion and $900 billion in 2026.

Bloomberg / Getty Images

The productivity measurement problem that nobody is solving yet

The productivity issue isn’t budget overruns. It’s measurement. AI is accelerating coding, experimentation, and workflows across departments, but companies still struggle to prove that speed translates into better products, faster releases, or higher profits.

Writer research shows that 79% of organizations report individual productivity gains from AI. Yet gains at the employee level don’t automatically translate into measurable business returns.

As Uber CFO Prashanth Mahendra-Rajah noted, the long-term impact remains difficult to quantify, even as underlying metrics improve dramatically, according to the Los Angeles Times.

Meanwhile, hyperscalers are investing heavily in AI infrastructure, with spending expected between $700 billion and $900 billion in 2026, according to Forbes.

The real challenge lies on the demand side: enterprises are still figuring out exactly what value they are getting from their AI investments.

Uber’s underlying business is strong even as the AI cost question hangs over it

The AI spending debate exists against the backdrop of a business that delivered a genuinely strong first quarter of 2026:

Gross bookings of $53.7 billion, up 25% year over year.
Non-GAAP operating income of $1.9 billion, up 42% year over year.
Non-GAAP EPS of $0.72, up 44% year over year.
Free cash flow of $2.3 billion.
3.6 billion trips in the quarter, up 20% year over year.
Uber One membership reaching 50 million, driving half of gross bookings. Source: Uber First Quarter 2026 Results

Q2 2026 guidance calls for gross bookings of $56.25 billion to $57.75 billion, representing 18% to 22% year-over-year growth on a constant-currency basis, with non-GAAP EPS of $0.78 to $0.82.

UBER is down 12.26% year-to-date compared to the S&P 500‘s 10.35% gain, according to Yahoo Finance. That’s a disconnect from the operational performance that may partly reflect investor uncertainty about AI cost trajectories and the $1.25 billion Rivian robotaxi commitment announced alongside its restructuring.

The AI spending cap at Uber is a maturation story of a company moving from unconstrained AI experimentation to managed AI deployment, not actually a failure story as most may see it.

The more interesting question, which Macdonald acknowledged honestly, is whether the organizational ROI eventually becomes as visible as the individual productivity gains. That answer is still being written.