Chamath Palihapitiya said corporate AI spending may have grown to a level that CEOs and CFOs still do not fully recognize, warning that earnings per share could one day come in a few cents lower with management unable to explain the gap.
Speaking on CNBC’s Squawk Box on July 14, the Social Capital founder, 8090 CEO and All-In Podcast host said internal AI usage spending, which he referred to as “tokenmaxxing,” has already reached a scale that many senior executives may not see.
Warning aimed at CEOs and CFOs
Palihapitiya said, “CEOs and CFOs, in my opinion, probably have no idea the extent of tokenmaxxing inside their organizations.”
He added that at some point, “you’ll see a company’s EPS suddenly miss by a few cents, and the reason for that gap will be underestimated AI spending.”
What “tokenmaxxing” refers to
According to the report, “tokenmaxxing” is a recently emerging Silicon Valley term describing an internal corporate policy of pushing employees to use AI as much as possible, based on the assumption that more AI consumption equals higher productivity.
Palihapitiya said that premise itself needs to be reconsidered. The report framed his comments as another public warning, following other prominent voices in the AI industry, that AI spending could end up hurting corporate profitability.
8090’s annual AI bill was nearing $10 million
Palihapitiya had also said publicly in March that annual AI-related spending at 8090, the AI advisory firm he founded, was approaching $10 million. He described that level of spending as “very scary” for a startup.
Reference to Jeremy Allaire’s paper
ABMedia also noted that on July 13, Chain News expanded on Circle CEO Jeremy Allaire’s paper The Agentic Economy. The sections titled “subscription death” and “models are cost, agents are business” discussed, at a structural level, a shift in AI consumption pricing from seats to units of work.