Last week a story went viral that I keep coming back to. Nicolas Brillante, head of strategic verticals at the fintech Slash, spent a few days using Claude to build a meme game called Brainrot Shooter. By the end of the week, the AI bill came to $81,267. One person. One week. His own words: "This was a genuine accident. I underestimated my own ability." He even called his shot, posting that he was about to become a case study for how AI spend gets out of control. So here we are.
The easy reaction is to laugh at the number. I think that misses the point. Slash had just told people to use AI coding more. Nicolas did exactly that. He had an unlimited key and an idea he kept chasing. Nobody told him to stop, because nobody could see the meter moving.
The part that makes it worse, and better
Here's the twist. The game actually took off. It went viral alongside the bill, people showed up to play, and Slash leaned in, joking that everyone should go play it so they could write the whole thing off as marketing. So the $81,267 might genuinely have been worth it.
That's the whole problem in one sentence. It might have been a great investment, and nobody could tell until the invoice arrived.
The villain is the dark room, not the developer
The villain here isn't Nicolas. It's the dark room he was working in. We have handed whole companies unlimited access to metered compute and wrapped almost no governance around it. No ceiling. No alert when one person's spend goes vertical. No way to glance at a number mid-week and decide "keep going" or "pull up." We gave everyone the keys and left the lights off.
And this is the part that should bother every leader: the same companies telling people to use more AI are the ones who will later demand to know who spent what. The people doing the spending were never handed a way to answer.
You can't govern what you can't see
The fix isn't to take the keys away. Brillante's instinct, build fast and chase the idea, was the right one, and it paid off. The fix is to turn the lights on. See the spend as it happens, attribute it to a person and a project, and put a ceiling where you actually want one. Do that, and an $81,267 week becomes a decision someone made on purpose instead of a surprise at the bottom of an invoice.
That's the reason we built TokenShift. Not to stop the next Nicolas. To make sure that when it happens, somebody actually chose it.
Unlimited access without visibility isn't empowerment. It's just an invoice you haven't read yet.