Y Combinator CEO Garry Tan says founders should spend $100,000 a year on AI tokens to access 2028-level capability today, even as Uber and Cognition push efficiency.
Alpha Score of 56 reflects moderate overall profile with weak momentum, moderate value, moderate quality, moderate sentiment.
Y Combinator CEO Garry Tan has a blunt message for founders second-guessing their AI spending: turn it up.
On Wednesday's A16z podcast, Tan said founders should load 800,000 or a million tokens into a single AI session without worrying about the bill. "You have to tune it all the way up," he said. "When you do that, I think that you basically get to live in 2028."
Tan's pitch is that heavy token consumption today gives founders access to AI capability most others will only see years from now. That advantage comes with a price tag. He put the cost at $50,000 to $100,000 a year for using agents at full strength.
"It just costs like a crazy amount," Tan said. "For a CEO or for a founder, it actually makes a lot of sense to do that."
Once an agent finishes a task, Tan said, founders should "skillify it" – turn the successful process into reusable instructions stored in a markdown file. He called that file "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 openly promoting tokenmaxxing – the practice of letting AI agents burn through large numbers of tokens rather than minimizing usage. Other voices in Silicon Valley have dismissed it as wasteful.
Uber's tech chief, Praveen Neppalli Naga, said last week that the ride-hailing company sees data suggesting the tokenmaxxing era is ending. "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 wrote on X.
Cognition CEO Scott Wu, whose company built the AI coding agent Devin, said in a June podcast that some companies have taken tokenmaxxing too far. "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," Wu said. He called tokenmaxxing "directionally correct" but said some have "gotten carried away."
The split reflects a real cost tradeoff. For a startup burning $100,000 a year on tokens, the bet is that speed today beats efficiency tomorrow. For larger companies like Uber, the math may favor optimization. Tan's advice is aimed squarely at the first group.
"A markdown file is an employee," he said. "It's an employee that will do the job perfectly every single time."
Drafted by a large language model from the source reporting linked above, then screened by automated publishing checks. It is not read by a journalist before publication. Some articles cite our Alpha Score. Verify prices and figures against the original source. Educational coverage, not personalized advice.