A meme game for $81,000: how a Slash employee accidentally spent a week's budget on AI tokens
Fintech startup Slash, valued at $1.4 billion, received an unexpected lesson in managing artificial intelligence expenses. One of its employees spent over $81,000 on AI tokens in a single week while creating a meme game. The trigger was a corporate directive to more actively use AI for writing code.
Nicolas Briante, head of strategic verticals, took this call more than literally. He dedicated an entire day to working with the Claude model, creating a shooter game called Brainrot Shooter. The game is filled with meme characters like Skibidi Toilet and Tung Tung Tung Sahur. The problem is that each request to the model, especially when uploading the full context of the codebase, consumes tokens. After a day of active development, the total exceeded tens of thousands of dollars.
Briante himself called the incident a "real accident." He simply underestimated how quickly AI consumption accumulates with repeated context uploads. The company reacted with humor: in a post on X, startup representatives joked that employees should play the game so it could be written off as marketing expenses. However, Slash is now reviewing its policy on using AI for writing code.
From Incident to Strategy
The story took an unexpected turn. In the first 48 hours, 6,912 people played Brainrot Shooter, with a total playtime of 8,986 hours and an average session of 1.3 hours. The peak number of concurrent gamers reached 437. The company received three incoming requests for advertising placements. The finance department reclassified the project from an expense incident to a strategic initiative.
Slash is not the only company to have encountered this problem in practice. Earlier this year, Uber exhausted its annual AI budget in four months, after which it introduced its own limits. And one unnamed company received a $500 million bill for a month of using Claude from Anthropic without setting restrictions for employees.
Expert opinion: This case is a vivid illustration of how a lack of control over AI spending can lead to unexpected financial consequences. The market for tokens and API access to models is still forming, and companies need to implement monitoring and limit systems; otherwise, "accidents" like Brainrot Shooter will become the norm. For the crypto industry, this is also a signal: tokens as a unit of accounting for AI computations require more transparent pricing and management mechanisms.