Copy article

Uber's COO admits its AI spending has produced no measurable increase in useful features

ended 27. May 2026

Uber has admitted its AI spending has produced no measurable increase in useful features, and its response was not to reduce AI spending.

Uber's COO Andrew Macdonald told the Rapid Response podcast that the company blew through its entire 2026 Claude Code budget by mid-March, with 25% of code commits now AI-generated. When he asked senior engineering leaders how many shelved projects had been rescued by those productivity gains, the answer was: none they could point to. The link between token consumption and consumer value, he said, is "not there yet."

Next, rather than cutting AI spending, Uber slowed hiring. Forbes reported the company's monthly AI spending per engineer ranged between $150 and $250 on average, while heavy users spent between $500 and $2,000 across roughly 5,000 engineers. 

CTO Praveen Neppalli Naga reportedly said he personally spent $1,200 during a two-hour demo session.

The cost of unproven AI tooling is being offset not by reducing that tooling, but by reducing people.

We'd like your views:

  • Uber burned through its annual AI code budget in under four months. Could that happen to other companies?
  • Macdonald says AI "seems free" to the individual developer but the company foots the bill. Should UK businesses be required to disclose AI tooling costs alongside headcount reductions in annual reports?
  • If UK firms bake AI adoption into performance reviews, does that create a de facto requirement to use tools regardless of fitness for purpose (tokenmaxxing)?
  • Uber's response to overspending on AI was to slow hiring, not to cut AI spend. Was that sensible?

5 responses from the Newspage community

Copy all

Star Quote
Copy

Colette MasonVerified
AI Ethics Consultant at Clever Clogs AI
Spending is only half the Uber story. The other half is what happens to the people still in the building when the budget goes on tools instead of them.

Duolingo found out the hard way. Its staff started asking whether they were being measured on outcomes or on obedience to a tool, and the company rethought its strategy.

Uber went the other direction: tokenmaxxing leaderboards, hiring slowed to fund the overspend, and a COO publicly admitting there's no visible link between any of it and a single new feature users actually wanted.

The people who survive these cuts know exactly what happened. They watched colleagues go while the tool that couldn't prove its value kept its seat. That does something to a workforce that no engagement survey captures and no meeting fixes.

If UK businesses are following this playbook, and the pressure to adopt is intense, they should be asking what it costs when your best people left conclude that the company values the tool more than the team.
Copy

Editor
Kate UnderwoodVerified
Founder & Chief People Strategist at Kate Underwood HR and Training
Uber burned through its entire 2026 AI coding budget by March. 25% of code commits are now AI generated. When leadership asked how many shelved projects had been rescued by those gains, the answer was none they could point to. Their response was not to cut AI spend. It was to slow hiring.
The cost of unproven tooling is being offset not by reducing that tooling, but by reducing people. That is a decision with a human price tag that never shows up in the AI budget line. When you cannot draw a straight line between spend and outcome, that is not a technology problem. That is a leadership problem. Spending £1,200 in a two hour demo session is not innovation. It is expensive curiosity with someone else's money. AI has enormous potential. But using real jobs as collateral while you wait for that potential to materialise is not a strategy. It is a wish list with a Gantt chart stapled to it.
Copy

Editor
Mitali DeypurkaysthaVerified
AI Governance Strategist & CMO at AISensa
UK firms are claiming AI productivity gains of 11.5%, according to Morgan Stanley research, yet Uber has just admitted it burned through its AI budget within months without pointing to any meaningful new features users actually wanted. That is starting to look less like innovation and more like the Emperor’s New Clothes for the AI age, where companies are so terrified of looking “behind” that they keep feeding the machine even when the return on investment is blurry. The worrying part is that when the numbers stop adding up, AI spending often remains protected, while the pressure falls on hiring and headcount. If businesses are going to freeze recruitment or cut jobs to bankroll experimental AI tools, shareholders and workers should be able to see those costs clearly rather than watching them disappear into a vague technology budget.
Copy

Editor
Samuel Mather-HolgateVerified
Managing Director & IFA at Mather and Murray Financial
With AI still in it's infancy, and lots of agents still producing hallucinations, companies are spending big amounts on AI at the moment, and not seeing the desired results. That said, in a tech area showing exponential growth and development, it wont be long before efficiencies are realised so the spending does still need to happen now.
Copy

Editor
Katrina YoungVerified
Futurist | Chief Technology Officer CTO/CDO Ethical AI & Digital Transformation Strategist at KYC Digital
Uber burning through its AI coding budget in under four months is not just a Uber problem. It is a warning about how many firms are measuring AI activity instead of business value. 25% AI-generated commits sounds impressive until leadership asks which products improved, which costs fell, or which projects were actually rescued. Token consumption is not productivity. Code volume is not progress. The governance problem appears when AI adoption metrics mature faster than operational measurement. Usage becomes visible before outcomes do. The real risk is not overspending. It is slowing hiring or reducing headcount to protect tooling budgets that still cannot prove measurable value. If organisations cannot clearly show outcomes delivered per pound of AI spend, the strategy is not mature yet. It is procurement wearing an innovation badge.