An AI startup accidentally burned through half a billion dollars in just one month on an artificial intelligence (AI) platform after forgetting to set usage limits for its staff. The massive $500 million bill, tied to the widespread use of Anthropic’s “Claude” AI by employees, was revealed in a new Axios report detailing a massive shift in corporate America. This shows that after a year of encouraging workers to experiment freely with AI, top executives are reportedly facing severe budget contraints and are suddenly forcing a reality check on soaring tech costs.
The Rise of ‘Tokenmaxxing’
Until recently, corporates believed AI was “the more, the better” as companies spent lavishly on AI subscriptions to prove to Wall Street that they wouldn’t be left behind. This corporate culture gave rise to a trend known as “tokenmaxxing” – where employees deliberately consumed as much computing power as possible just to appear tech-forward.Since early AI pricing models were heavily subsidised, companies didn’t notice the dent. However, now that top AI developers have shifted to usage-based pricing, corporations are paying for every single “token” (the basic unit of measurement for AI computing) that their employees consume, as per a report by The Wall Street Journal.
What Meta , Uber, Salesforce and other tech companies have said
The financial fallout has been swift with companies like Uber, Instagram, Facebook and WhatsApp-parent company Meta, Salesforce and others hinting at judicious usage. Google revealed that its token processing has skyrocketed to 3.2 quadrillion tokens a month, which is seven times higher than last year.
- Uber blew through its entire annual budget for autonomous AI tools in just three months.
- Factory, a tech automation firm, reported that employees at a top financial institution were burning through hundreds of thousands of dollars a month using ultra-expensive AI subscriptions for casual small talks or answer basic questions.
- Meta CTO Andrew Bosworth issued a blunt warning to employees in a recent internal memo: “Nobody should be using AI tools just for the sake of using them. All motion is not progress, and token usage alone is not a measure of impact of any kind,” Bosworth wrote.
Other major players are taking immediate action to rein in the waste:
- Salesforce has built an internal tracking system to ensure that every AI token used by a worker directly contributes to a profitable business outcome.
- Microsoft also stripped away employee access to Anthropic’s Claude programs, choosing instead to standardise operations and push workers toward internal, homegrown coding assistants.
- According to a Financial Times report, Amazon has shut down its internal leaderboard that tracked employees’ use of Al.