Y Combinator’s CEO has a message for founders worried about burning too many AI tokens: Burn, baby, burn.
On an episode of the A16z podcast released on Wednesday, Garry Tan said that founders should give themselves permission to spend heavily on AI tokens — even if it comes with a hefty price tag.
“You have to tune it all the way up. You’re just, like, let me load a million tokens or 800,000 tokens in,” he said about using AI agents. “When you do that, I think that you basically get to live in 2028.”
Tan’s pitch is that spending heavily on AI agents lets founders access a level of AI capability most others will only have access to in the future.
He added that this advantage won’t come cheap.
“It costs, I don’t know, 50 or $100,000 a year to use the agents at full strength,” he said. “It just costs like a crazy amount. But for a CEO or for a founder, it actually makes a lot of sense to do that.”
Tan said that founders can make that expensive experimentation pay off over time. Once an agent successfully completes a task, he said, founders should “skillify it” by turning the process into reusable instructions that the AI can follow automatically again and again.
“A markdown file is an employee,” he said, referring to a written set of instructions that tells an AI agent how to perform a particular job. “It’s an employee that will do the job perfectly every single time, and it’ll do it as many times as you want.”
Tan is a rare tech leader championing tokenmaxxing, even as other voices in Silicon Valley have dismissed it as wasteful. Tokenmaxxing is the practice of deliberately letting AI agents consume large numbers of tokens rather than minimizing their use and cost.
Last week, Uber’s tech chief, Praveen Neppalli Naga, said the ride-hailing company is seeing interesting data on AI costs and that this was “another signal that we’re coming to the end of the so-called ‘tokenmaxxing’ era.”
“The next phase, whatever we call it, will not be characterized by who spends the most tokens, but about how people use them as efficiently as possible,” he said in a post on X.
In a June podcast, AI coding giant Cognition’s CEO Scott Wu said companies lost the plot by encouraging tokenmaxxing.
“It is directionally correct, but I think there are definitely some places where people have gotten carried away,” Wu said. “People are like, ‘We rank our engineers by how many tokens they’re spending.’ Well, let’s try and rank people by how much output they’re actually producing.”
