Asakiyume mita

Posted by David Gerard

https://pivot-to-ai.com/2026/04/28/tokenmaxxing-how-much-did-you-spend-in-tokens-ceo-of-tokens/

https://pivot-to-ai.com/?p=7868

Let’s just marvel at the AI jargon word “toxenmaxxing” — when you’re AI coding at work, run up as absolutely high an AI token bill as you possibly can. Why? Because it’s the future! And not just stupid.

The idea’s been hot with the AI bros for a while now. But tokenmaxxing really joined the chat when Jensen Huang of Nvidia went on the All-In Podcast on 19 March: [YouTube]

Let me give you a thought experiment. Let’s say you have a software engineer or AI researcher and you pay them $500,000 a year. We do that all the time, okay, this is happening all of the time. That $500,000 engineer, at the end of the year I’m going to ask him how much did you spend in tokens, and that person says $5,000, I will go ape-something-else. If that $500,000 engineer did not consume at least $250,000 worth of tokens, I am going to be deeply alarmed, okay, and this is no different than one of our chip designers who says, guess what, I’m just going to use paper and pencil, I don’t think I’m going to need any CAD tools.

Jensen sells the cards the tokens run on. Anthropic lose several dollars on every dollar they make, but Jensen makes money when you burn out a few more Nvidia cards. This is the CEO of tokens telling you to spend more on tokens.

On April 6th, The Information discussed Meta’s “Claudeonomics” leader board: [Information, archive]

Employees at Meta Platforms who want to show off their AI superuser chops are competing on an internal leaderboard for status as a “Session Immortal”—or, even better, “Token Legend.

… The practice is emblematic of Silicon Valley’s newest form of conspicuous consumption, known as “tokenmaxxing,” which has turned token usage into a benchmark for productivity and a competitive measure of who is most AI native.

CEOs and CTOs knew that AI shook off expensive whiners who pushed back on AI. Tokenmaxxing would shake loose these AI blockers, these saboteurs! The way out is the way through!

Here’s Sonya Huang, a venture capitalist at Sequoia Capital: [WSJ, archive]

Critics of tokenmaxxing are missing the point, she said. Yes, it’s an imperfect metric, she admitted, but “the thing that matters for your company is: is my employee becoming insanely AI-pilled? And that requires getting them on this tokenmaxxing mindset.”

She literally doesn’t know if it’s even working. So do it harder!

Anthropic’s turning the screws on customers about pricing. But the power of making you spend all your money on tokens is to make token spending capability the new moat around using AI coding at all. You can only compete if you’re huge — or if you’re blowing venture capital cash as fast as possible.

That works fine — except the bit where AI code is sort of terrible. Every time prominent AI code leaks, we see it’s rambling, ill-constructed, and full of bugs and security holes.

The main product of AI coding is code churn: [TechCrunch]

The data from across the industry tells a consistent story: More code is being written, but a disproportionate amount of it isn’t sticking.

Multiple surveys show code churn rates are eight to nine times or more what they were before AI everywhere. Slop coders keep having to do stuff over. Because the code is rubbish. [GitClear, PDF; Faros]

At some point, Anthropic puts up its prices so much the companies try to go back to brain coders. They hand these actual human coders a toxic waste dump of AI slop code that is literally incomprehensible to humans and can’t be fixed in less time than it would take to start over. This does not end well.

But that sounds like a next year problem. It’s still this year! Use more tokens! Business doesn’t exist to do things. Use more tokens.

• Video — Podcast

It’s pledge week at Pivot to AI! If Pivot brightens your day, please do put $5 into the Patreon. Tell your friends!

https://pivot-to-ai.com/2026/04/28/tokenmaxxing-how-much-did-you-spend-in-tokens-ceo-of-tokens/

https://pivot-to-ai.com/?p=7868