Uber and Tokenmaxxing: A Frantic Race Without Tangible Gains

Uber and the Skepticism Surrounding Tokenmaxxing

Andrew Macdonald, Uber's Chief Operating Officer, recently expressed his reservations about the effectiveness of AI tokens in improving productivity. In an interview, he highlighted the lack of concrete evidence linking the increased use of these tokens to a significant rise in useful features for consumers. "That link isn't there yet, is it?" he stated, remarks that quickly garnered attention on social media, reaching over 2 million views on X.

AI tokens, which represent about ¾ of a word each, are essential for the functioning of AI chatbots. The concept of "tokenmaxxing" involves maximizing the use of these tokens to boost productivity, although tangible results are still under debate.

The Massive Adoption of AI by American Companies

Many companies in the United States, such as Meta, Disney, and JPMorgan, are aggressively integrating AI. Meta has even rebranded some of its employees as "AI builders," encouraging them to work in AI-native "pods." Visa, on its part, has highlighted its monthly spending on tokens, which approaches 2 trillion, while rewarding teams that build faster with AI.

However, this rapid adoption is not without criticism. Many tech industry professionals believe it leads to massive resource waste. Uber, for example, has already exhausted its AI budget for the year in just four months, as reported by the Information.

Growing Concerns About AI Spending

Akshat Bubna, co-founder of the AI startup Modal, expressed his concerns on X, stating, "I'm almost sure that 50% of internal token spending is completely unnecessary, but for now, it's hard to know which 50%." Karthik Hariharan, a technical lead, also pointed out the lack of significant return on investment despite millions spent.

Sundar Pichai, CEO of Google, recently stated at Google’s flagship developer conference, I/O, that Chief Information Officers are worried about rising AI spending, anticipating that the problem will worsen over the year. Meanwhile, Michael Burry, known for his financial predictions, referred to tokenmaxxing as a "crazy phase" and warned of a potential drop in Nvidia's stock.

Towards a More Balanced Use of AI Tokens

Despite the criticism, some defend tokenmaxxing. Garry Tan, CEO of Y Combinator, claimed that his company has been practicing this strategy for a long time. A report from Jellyfish suggests that a balanced use of AI is possible. The report revealed that the top 10% of Claude Code users consumed about 10 times more AI tokens than the median developer and produced only about twice as many results. It recommends linking costs to concrete metrics, such as pull requests, rather than simply rewarding raw token consumption.