Blab Tech
Chamath Palihapitiya Warns AI Token Spending Could Hit Corporate Earnings
When venture capitalist Chamath Palihapitiya warned that corporate earnings could be blindsided by hidden AI token costs, the message was clear: the invisible expense is growing.
On Tuesday, Palihapitiya—known for Social Capital and the AI startup 8090—told reporters that many CEOs and CFOs may not realize how much token‑maxxing is happening inside their organizations. According to CNBC, he said executives “probably have no idea how much tokenmaxxing is going on inside of their organizations.” He added that a future earnings‑per‑share miss could prompt a CEO to ask a CFO, “What happened?”
Palihapitiya’s remarks arrive on the heels of a series of public comments about the cost of AI. In March, he posted on X that his own company’s AI spend was trending toward more than $10 million a year—a figure he described as “feels very scary” for a small startup. He said the token‑maxxing trend—where employees are encouraged to consume as many AI tokens as possible—was ending.
The term token‑maxxing refers to maximizing token usage as a proxy for productivity. The practice has been discussed in industry blogs and forums, and some analysts say the focus is shifting toward value‑maxxing, which emphasizes tangible outcomes over raw token consumption.
Palihapitiya’s concerns echo those of Palantir CEO Alex Karp. Karp told CNBC’s “Squawk Box” that he was not “throwing shade” at OpenAI or Anthropic, but that “something has gone completely wrong.” He added that many enterprises in the United States are “chillaxing and wasting time with tokens.” Karp’s comments were reported by Yahoo Finance and CryptoBriefing.
The AI spending issue is part of a broader conversation about the economics of large‑language‑model (LLM) services. According to reports, the cost of token usage has risen sharply, while the productivity gains reported by some companies have not kept pace. The discrepancy has prompted executives to question the return on investment for AI initiatives.
Palihapitiya’s own background in the AI and venture space may explain his focus on token economics. He founded 8090 in 2024, a platform that lets teams collaborate with AI agents to build enterprise software. In June, 8090 announced a $135 million funding round led by Salesforce. The company’s CEO, Palihapitiya, said the startup’s spend on AI tokens was growing rapidly.
Beyond AI, Palihapitiya has been a prominent figure in the special‑purpose acquisition company (SPAC) market. He launched the American Exceptionalism Acquisition Corp. A (AEXA) last year, a SPAC that targets companies in AI, energy, defense, and decentralized finance. Palihapitiya has been criticized for his role in promoting SPACs during the COVID‑19 pandemic, many of which have since shuttered and resulted in losses for investors. He has admitted that “who didn’t make money? Speculators,” and that it was a “huge mistake” to promote SPACs on social media and CNBC.
Palihapitiya’s comments about AI token spending are not the first time he has warned about hidden costs. In a 2026 CNBC interview, he said that executives may be blindsided by unseen token usage, leading to earnings misses. He also noted that the era of token‑maxxing is over, a view that aligns with the broader industry shift toward measuring real value.
The discussion highlights a tension between the rapid adoption of AI services and the need for clear cost controls. Companies that have embraced token‑based pricing models may need to reassess how they track usage and link it to business outcomes. Investors and regulators are watching closely, as the AI boom continues to reshape enterprise spending.
At present, Palihapitiya’s 8090 remains a private company, and no public financial statements are available. The company’s $135 million round was led by Salesforce, but the terms of the deal have not been disclosed. Palihapitiya’s AEXA is listed on the NYSE, but its target acquisition has not yet been announced.
The conversation about AI token economics is likely to continue as more companies report their spend and as the industry seeks ways to align AI usage with measurable returns. For now, executives are being cautioned to monitor token consumption closely to avoid unexpected earnings impacts.
On Tuesday, Palihapitiya—known for Social Capital and the AI startup 8090—told reporters that many CEOs and CFOs may not realize how much token‑maxxing is happening inside their organizations. According to CNBC, he said executives “probably have no idea how much tokenmaxxing is going on inside of their organizations.” He added that a future earnings‑per‑share miss could prompt a CEO to ask a CFO, “What happened?”
Palihapitiya’s remarks arrive on the heels of a series of public comments about the cost of AI. In March, he posted on X that his own company’s AI spend was trending toward more than $10 million a year—a figure he described as “feels very scary” for a small startup. He said the token‑maxxing trend—where employees are encouraged to consume as many AI tokens as possible—was ending.
The term token‑maxxing refers to maximizing token usage as a proxy for productivity. The practice has been discussed in industry blogs and forums, and some analysts say the focus is shifting toward value‑maxxing, which emphasizes tangible outcomes over raw token consumption.
Palihapitiya’s concerns echo those of Palantir CEO Alex Karp. Karp told CNBC’s “Squawk Box” that he was not “throwing shade” at OpenAI or Anthropic, but that “something has gone completely wrong.” He added that many enterprises in the United States are “chillaxing and wasting time with tokens.” Karp’s comments were reported by Yahoo Finance and CryptoBriefing.
The AI spending issue is part of a broader conversation about the economics of large‑language‑model (LLM) services. According to reports, the cost of token usage has risen sharply, while the productivity gains reported by some companies have not kept pace. The discrepancy has prompted executives to question the return on investment for AI initiatives.
Palihapitiya’s own background in the AI and venture space may explain his focus on token economics. He founded 8090 in 2024, a platform that lets teams collaborate with AI agents to build enterprise software. In June, 8090 announced a $135 million funding round led by Salesforce. The company’s CEO, Palihapitiya, said the startup’s spend on AI tokens was growing rapidly.
Beyond AI, Palihapitiya has been a prominent figure in the special‑purpose acquisition company (SPAC) market. He launched the American Exceptionalism Acquisition Corp. A (AEXA) last year, a SPAC that targets companies in AI, energy, defense, and decentralized finance. Palihapitiya has been criticized for his role in promoting SPACs during the COVID‑19 pandemic, many of which have since shuttered and resulted in losses for investors. He has admitted that “who didn’t make money? Speculators,” and that it was a “huge mistake” to promote SPACs on social media and CNBC.
Palihapitiya’s comments about AI token spending are not the first time he has warned about hidden costs. In a 2026 CNBC interview, he said that executives may be blindsided by unseen token usage, leading to earnings misses. He also noted that the era of token‑maxxing is over, a view that aligns with the broader industry shift toward measuring real value.
The discussion highlights a tension between the rapid adoption of AI services and the need for clear cost controls. Companies that have embraced token‑based pricing models may need to reassess how they track usage and link it to business outcomes. Investors and regulators are watching closely, as the AI boom continues to reshape enterprise spending.
At present, Palihapitiya’s 8090 remains a private company, and no public financial statements are available. The company’s $135 million round was led by Salesforce, but the terms of the deal have not been disclosed. Palihapitiya’s AEXA is listed on the NYSE, but its target acquisition has not yet been announced.
The conversation about AI token economics is likely to continue as more companies report their spend and as the industry seeks ways to align AI usage with measurable returns. For now, executives are being cautioned to monitor token consumption closely to avoid unexpected earnings impacts.
