Chamath Palihapitiya Says Tokenmaxxing Weighs on Earnings — ‘CEOs and CFOs Have No Idea How Much Is Going on Inside Their Organizations’

Artificial intelligence has become the office’s newest power tool. But when every employee starts reaching for it, someone eventually has to pay the bill.

Venture capitalist and Social Capital founder Chamath Palihapitiya believes many companies are about to discover just how expensive that bill has become. Speaking to CNBC on July 14, he said executives may have far less visibility into AI spending than they think, thanks to a growing trend known as “tokenmaxxing”—the heavy use of AI models across an organization.

“CEOs and the CFOs, in my opinion, probably have no idea how much tokenmaxxing is going on inside of their organizations,” Palihapitiya said. “I suspect what’ll happen is one day you’re going to have a miss, and EPS will be off by a few pennies, and the CEO will say to the CFO, ‘What happened?’”

When AI Costs Start Outrunning the Payoff

Palihapitiya isn’t speaking only as an investor. He is also CEO of 8090, an AI-native enterprise software company he founded in 2024. The startup, which raised a $135 million Series A funding round in June, builds AI software for industries including healthcare, financial services, and manufacturing.

Read more: Apple just told us where the next decade of tech is headed

Read more: The next AI wave won’t look like chatbots. It will look like this.

His latest comments build directly on remarks he made during the July 11 episode of the All-In Podcast (which he co-hosts), where he shared real-time insights from inside his own company, 8090.

“I sat down with my CTO today, and I said, ‘How are we doing on token spend?’ He said the most incredible thing. He said, ‘Right now, our token costs are doubling every 45 days,’” Palihapitiya said. “And I said, well, what is the downstream productivity? And he said, ‘Maybe 5% max.’”

“So my costs are doubling every 45 days; my upside is essentially flat,” he continued, adding that achieving the next round of improvements requires dramatically more AI usage than before.

Earlier this year, he also revealed that 8090’s annual AI bill was on pace to exceed $10 million, a figure he described as “very scary” for a company of its size.

So, What Exactly Is Tokenmaxxing?

The name may sound like internet slang, but the idea is relatively simple.

AI companies such as OpenAI and Anthropic charge businesses based largely on “tokens,” the pieces of text their models process every time someone submits a prompt or receives a response. The more employees use AI, the more tokens a company consumes—and the higher the bill.

Tokenmaxxing describes the growing habit of encouraging widespread AI use without closely measuring whether the productivity gains justify the rising costs.

For many businesses, AI adoption happened organically as employees found new uses for chatbots, coding assistants, and automation tools. Over time, those small expenses can add up to a meaningful operating cost.

The Next AI Test Could Be Earnings Season

Palihapitiya’s broader point is that AI spending is entering a new phase.

Early adopters often saw quick wins through tasks like summarizing documents, generating code, and automating repetitive work. But as companies ask AI to tackle more sophisticated problems, costs can rise much faster than productivity.

If finance teams aren’t closely tracking those expenses, the impact may not become obvious until quarterly earnings reveal shrinking margins or weaker-than-expected profits.

That could also push companies to rethink their AI strategies by focusing less on using as much AI as possible and more on whether each dollar spent is producing measurable returns.

For Palihapitiya, the next chapter of the AI boom won’t be defined by who uses the most tokens. It will be defined by who can prove they’re worth paying for.