- In Y Combinator's "Startup School," partner Diana Hu advocated for startups to begin tokenmaxxing.
- "Maximizing token usage, not head count, will be the critical shift," Hu said.
- Tokens measure spending on AI compute. Hu said founders need to be willing to run an "uncomfortably high API bill."
Tokenmaxxing isn't just a trend; it's bona fide advice from Silicon Valley's leading startup accelerator.
In a new episode of Y Combinator's "Startup School," partner Diana Hu instructs founders how to build an AI-native company. Hu was a YC-backed founder herself, building augmented reality company Escher Reality, which was acquired by Niantic.
"Maximizing token usage, not head count, will be the critical shift," Hu said. "The best companies will be the ones that are tokenmaxxing."
Tokens measure the cost of AI computing. The more tokens spent, the more an individual employee or developer has used their AI tools. (Importantly, more tokens doesn't necessarily more impact.) Some companies have built token leaderboards or incentivized tokenmaxxing, the process of spending as many tokens as possible.
Business Insider asked startup leaders about the trend. Some said that tokenmaxxing was a no-brainer; others said that it didn't make sense at their size.
Hu, like her boss Garry Tan, is an unabashed proponent. She described the "tradeoff" between labor and token spending.
"One person with AI tools can be the equivalent of what used to take a large engineering team at a pre-AI company," Hu said. "That means dramatically leaner engineering, design, HR, and admin teams."
Startup founders should "be willing to run an uncomfortably high API bill because it's replacing what would have taken a far more expensive and inflated head count," Hu said.
While Y Combinator's advice to startups may not translate to larger companies (though there are plenty of VCs and voices in the tech world that would argue it should), the instructional video offers an interesting look at the operational values being instilled in the next generation of up-and-coming CEOs.
Hu also voiced support for a three-pronged employee base. There are individual contributors (who build things), the directly responsible individual (who focuses on strategy), and the AI founder (who leads while still building).
It's a similar structure to Jack Dorsey's redesign of Block, his payment processing company. After laying off about 40% of staff, Dorsey announced a new, three-pronged structure to turn Block into a "mini-AGI."
Founders must try the tools for themselves, too. Hu said that leaders should not "outsource" their belief in these AI tools.
"You need to develop it yourself by actually sitting with coding agents and using them until you start to break your own priors about what is now possible to build," Hu said.
