Uber raced through an entire year's worth of AI spending in just a few months — and its own chief technology officer now admits the era of unchecked AI spending is over.
In an interview with The Information earlier this year, Uber CTO Praveen Neppalli Naga acknowledged he had to go "back to the drawing board" on AI budgets after the rideshare giant's aggressive internal push burned through its 2026 allocation almost immediately.
How Uber's AI spending spree went off track
Uber had told employees — particularly software engineers — to use AI tools as much as possible. The push centred on Anthropic's Claude Code, an AI coding assistant, and went as far as creating internal "leaderboards" that ranked engineers by how heavily they used the tool.
The strategy worked almost too well. Usage surged, costs followed, and the year's AI budget evaporated within months.
What 'tokenmaxxing' actually means
"Tokenmaxxing" describes companies actively incentivising employees to maximise their use of AI tools — often through rewards, leaderboards, or making AI adoption part of performance expectations.
The logic appeared straightforward: more AI usage should mean higher productivity. But many companies discovered the returns did not justify the relentless spending, and quietly pulled back.
What the CTO's admission signals to the industry
According to the report, Uber was no exception to this broader retreat. Naga's statement that "we're coming to the end of the so-called tokenmaxxing era" marks a striking reversal from the company's earlier enthusiasm.
His comment reflects a growing recognition across the tech industry that raw AI consumption is not the same as AI value.
Why this matters for companies betting big on AI
Uber's experience shows how quickly AI costs can spiral when adoption is pushed indiscriminately across a large workforce.
For CFOs and technology leaders watching from India and elsewhere, the lesson is blunt: usage-driven AI incentives can generate enthusiasm, but without clear return-on-investment guardrails, they can consume budgets at an alarming pace.
What remains unclear about Uber's AI budget
Exactly how much Uber spent, which cost components overwhelmed the budget, and what the company's revised AI spending plan looks like have not been disclosed in available reporting. The full details of Naga's conversation with The Information also remain unpublished.
Risks, criticism, and the balanced view
Not everyone considers tokenmaxxing a failed experiment. Supporters argue that early, aggressive AI adoption builds important organisational capability — even when initial costs are steep.
Critics counter that metrics like leaderboards reward volume over value, encouraging AI usage for its own sake rather than for meaningful productivity gains. The truth, as Uber's experience suggests, lies somewhere in between.
The wider shift in enterprise AI adoption
Uber is not alone. Across the technology sector, companies that once celebrated raw AI engagement are now tightening budgets and asking harder questions about measurable outcomes.
The "tokenmaxxing era" may be ending, but a far more demanding phase — one focused on accountability and demonstrated returns — is just beginning.
What should companies do now
Organisations pushing AI adoption should define clear success metrics before scaling usage incentives. Track cost per meaningful outcome, not just tokens consumed or tool logins.
For engineers and employees, the shift means AI skills still matter — but demonstrating how AI improves actual work output will matter far more than usage volume.
Our Take
The most telling detail in this story is not the budget overrun itself, but its speed. A full year of AI spending gone in months is a vivid illustration of how quickly enthusiasm can outpace discipline.
Naga's "back to the drawing board" comment suggests Uber is not abandoning AI — it is recalibrating. A company that once ranked its engineers on AI usage is now learning to rank AI spending on returns. If that discipline spreads, it could define the next phase of enterprise AI.
Frequently Asked Questions
What is tokenmaxxing?
Tokenmaxxing is a workplace trend where companies incentivise employees to maximise their use of AI tools, sometimes through leaderboards or usage-based rewards, to accelerate AI adoption across the organisation.
Why did Uber blow through its 2026 AI budget?
Uber encouraged employees, especially software engineers, to use AI tools like Anthropic's Claude Code as much as possible. The resulting surge in usage and associated costs exhausted the full-year AI budget within the first few months.
Who is Uber's chief technology officer?
Praveen Neppalli Naga is Uber's chief technology officer. He told The Information that he went "back to the drawing board" on AI spending after the budget overrun and said the tokenmaxxing era is ending.
Is Uber stopping its AI investments?
No. The reported comments suggest Uber is recalibrating its AI approach rather than halting it. The CTO's statement signals a shift from encouraging maximum AI usage toward more measured, returns-focused adoption.
