A forceful statement by Palantir's CEO, Alex Karp, has brought a long-simmering tension in the tech industry to the forefront: major AI labs are building momentum using client data and decisions, while traditional enterprises grow increasingly worried about being relegated to mere "value contributors" in this AI wave.
Over the past two weeks, Karp first delivered a nearly 20-minute fiery critique on CNBC, accusing AI labs of overstating capabilities, overpricing tokens, and claiming that every large enterprise client he speaks with is "furious" about it.
Subsequently, Palantir released a white paper titled "Institutional Sovereignty in the Age of AI," outlining 15 recommendations for businesses and governments to guard against the erosion of their core data by AI giants like OpenAI and Anthropic. These actions quickly sparked widespread discussion in tech circles.
At the heart of this debate is a question being asked more loudly: who captures the value in the AI era—the enterprises deploying AI, or the labs developing the underlying models?
This issue is not only about the commercial landscape but has also spread to the realms of policy competition and geopolitical rivalry, posing a direct threat to the valuations of traditional software providers.
Karp is Not a Lone Voice
Karp himself acknowledges his position is not neutral.
The core product of Palantir Technologies Inc. is a middle layer built on foundational models, connecting AI with enterprise clients, a positioning that gives the company a direct commercial stake in the tug-of-war between enterprises and AI labs.
Responding to skepticism that he is merely venting emotions, Karp stated that this is the voice of American business, channeled through him.
It is noteworthy that Karp is not the only tech executive sounding the alarm about this imbalance.
Microsoft CEO Satya Nadella has recently written and publicly expressed similar concerns, with his core worry being whether the "learnings" accumulated by a company using an AI model can truly be retained by the company itself.
Nadella said at a Stanford University event this month that if you are merely a consumer of a foundational model, he is unsure how you retain enterprise value, let alone create it.
The "Encroachment" Strategy of AI Labs
Karp's criticism has touched a deeper nerve of anxiety within the tech industry.
Former White House AI lead David Sacks quickly echoed this view on social media, directly targeting Anthropic. Sacks wrote that Anthropic has successively launched Claude Science, Claude Security, Claude Legal, and Claude Code—each product directly entering fields previously served by companies building applications on its model.
Sacks further stated that the pattern is consistent: observe where value is being created, then move in directly. Dominate the model layer first, then leverage that position to capture the most lucrative vertical markets.
This "observe, replicate, expand" path is unsettling for the many companies building commercial applications reliant on large model APIs. For these firms, contributing data and use cases to AI labs may be providing ammunition for a future competitor's entry.
Neither OpenAI nor Anthropic has publicly responded to Karp's criticism. Both companies' current policies state that enterprise client data is not used to train their models.
An AI lab insider dismissed the critique, suggesting it would be foolish to respond to Karp's performance, which is merely him advocating for his own interests.
Winners are Not Yet Decided
The deeper backdrop to this debate is the high level of uncertainty across the industry regarding where AI value will ultimately reside.
It was reported this week that Starbucks is using AI to replace software previously purchased from Microsoft and IBM, putting pressure on both companies' stock prices.
This case is seen as a microcosm of AI accelerating the reshaping of the enterprise software landscape. Analysts note that, in the time it takes to drink a coffee, winners and losers in the AI era could change, reinforcing the harsh reality that current tech giants are not guaranteed future leadership.
Meanwhile, Meta announced a new version of its AI model over the weekend and introduced a paid tier. According to a Bloomberg report, Meta CEO Mark Zuckerberg explicitly stated in an interview that he sees an opportunity to compete on price, as some other labs have very extreme pricing with extremely high margins, and he believes it is possible to offer state-of-the-art or high-level intelligent services at a more affordable cost.
This statement further intensifies competitive pressure in the foundational model market and lends credence to Karp's criticism that AI lab pricing is inflated.
Comments