Last month, Satya Nadella sounded the alarm for corporate America by questioning the true cost of enterprise AI.
The Microsoft (MSFT) CEO argued that businesses are essentially paying twice for using AI services: once for the intelligence and again for the proprietary knowledge they must give up to obtain it.
That covers everything from prompts and corrections to internal workflows, all of which can be included in that hidden second payment.
I covered that story, titled “Microsoft CEO adds fuel to Palantir CEO’s AI warning,” as that fire was originally kindled by Palantir CEO Alex Karp in a fiery CNBC interview.
Karp, of course, is no stranger to controversial claims.
However, during Palantir’s (PLTR) latest earnings call on August 3, he took the argument somewhere even more unsettling, just as the business was celebrating another blockbuster performance.
Despite the eye-catching headline numbers, Karp used that moment to raise a much bigger question about who captures the value created by corporate AI spending.
John Lamparski/Getty Images
Are companies paying AI labs to build their future competitors?
Alex Karp just took Satya Nadella’s critique of the AI business to another level.
More AI:
- Nvidia just made a move Wall Street wasn’t ready for
- Microsoft just took sides in AI policy fight
- OpenAI just disclosed something genuinely alarming
During Palantir’s Q2 earnings call, Karp argues that the ‘second payment’ Nadella talked about could eventually transfer part of a company’s competitive advantage to an outside AI provider that could replicate, commoditize, or compete with its business.
General-purpose models are usually starting points, and over time, interactions produce something much more valuable than the company’s original data.
For instance, it may include a detailed record of how its best employees reason, make trade-offs, and solve issues.
Karp described this provocatively. According to him, enterprises are paying model providers “to migrate your IP, your know-how, and your expertise to their model.”
So, the operational knowledge that separates one business from another is encoded in the outside provider’s broader tech stack.
That paints a troubling economic asymmetry.
The enterprise is for tokens, supplying context, corrections, and whatnot to improve output.
However, the outside AI provider ends up retaining the stronger long-term position, controlling the underlying model, infrastructure, and potentially the intelligence generated through ongoing use.
According to Karp, the eventual outcome is a provider building “a competitive business that doesn’t require your business or your people.”
Palantir Chief Revenue Officer Ryan Taylor sharpened the argument further, saying that businesses are “paying to give away their most important secrets.” Those secrets could become training material embedded in future models, contributing to the “commoditization of their own businesses.”
Karp refers to that accumulated advantage as a company’s alpha.
He believes businesses need to ensure that AI compounds that alpha internally instead of exporting them to a third party.
How much has Palantir stock gained post-earnings?
Palantir stock jumped almost 15% in after-hours trading on August 3, rising from its regular-session close of $125.65 to $144.45 following its Q2 earnings beat and guidance raise.
According to Seeking Alpha, over the past six months, the stock has tanked 21%, trailing the S&P 500’s 10% gain over the same period. Year-to-date, things get even uglier, with the stock down 30%.
It’s important to note, though, that over the years, Palantir stock has been a tremendous wealth creator, rising 570% over a three-year period.
How strong was Palantir’s quarter, really?
Palantir’s superb Q2 results showed that the company’s growth is accelerating, even as the business gets much bigger.
For perspective, after another stellar Q2 showing, Palantir has now sped past top-and-bottom-line estimates in each of the past four quarters.
Revenue surged to $1.935 billion, up 93% year over year and 19% sequentially, beating estimates by $130 million.
The U.S. business was the primary engine, growing 115% to $1.573 billion and representing 81% of total sales. U.S. commercial revenue led the charge, surging 149% to $764 million, while U.S. government revenue shot up 90% to $809 million.
That’s an impressive stat, which indicates that Palantir isn’t relying on one side of its business to carry the other. Commercial adoption is broadening at a rapid clip, while government demand remains unusually strong.
Moreover, commercial total contract value reached $2.337 billion, while U.S. commercial TCV jumped 153% to $2.132 billion, suggesting that revenue growth is driven by future commitments rather than a single exceptionally robust quarter.
A stand-out customer metric from the quarter was that revenue from Palantir’s top 20 customers jumped 67% to an average of $124 million each, which shows major clients are growing meaningfully after initial deployments.
Profitability also strengthened alongside growth.
Adjusted operating income reached $1.194 billion, producing a 62% margin, while adjusted free cash flow was $1.22 billion, equalling a 63% margin. That pushed Palantir’s Rule of 40 score to 155, which is exceptional even for a premium software business.
Palantir’s Rule of 40adds its sales growth rate to its adjusted operating margin; anything above 40% is generally considered impressive for a software business.
For instance, 48% revenue growth + 46% adjusted operating margin = 94% Rule of 40 score.
Additionally, management’s guidance implies confidence that the momentum can continue. Palantir bumped its full-year revenue guidance to roughly $8.154 billion, implying 82% growth, lifting its U.S. commercial growth outlook to at least 134%.
Q3 sales are expected to reach nearly $2.162 billion, with adjusted operating income near $1.294 billion.
There were a few caveats, though, that are worth talking about.
Adjusted expenses jumped 37%, stock-based compensation reached $265 million, and adjusted earnings per share got a considerable $0.02 lift from unrealized SpaceX (SPCX) gains. On top of that, gross margins also faced pressure after Palantir absorbed cloud-hosting costs for one government customer.
Could sovereign AI become Palantir’s biggest investor catalyst?
Palantir’s solution to the enterprise AI ownership conundrum is what Karp calls “sovereign AI.”
Put simply, customers retain control over their data, operating logic, security architecture, and model weights. At the same time, they preserve the flexibility to replace one model with another.
In that framework, the model effectively becomes more of a component rather than the core of the entire AI system.
Karp argues that models are fine-tuned inside a company’s own environment, using Palantir’s software and Nvidia (NVDA) infrastructure. Moreover, he claims they can outperform frontier models on specific business tasks without compelling customers to surrender the resulting intelligence.
“You own the weights, you own the alpha. You own everything,” he said.
For investors, that is the core of Palantir’s argument.
If enterprises continue to view dependence on closed AI providers as a competitive risk, spending would likely shift toward private deployments, open-weight models, model-switching tools, and orchestration platforms operating within corporate security boundaries.
The value shifts away from the model to the application layer, where Palantir continues to build a meaningful advantage.
Moreover, Karp was remarkably direct about what that opportunity entails for shareholders on the earnings call.
“For everyone on the sidelines, you got to get off the sidelines,” he said. He later told individual investors that Palantir was on its way to becoming a “much, much, much, much larger company.”
Moreover, Karp said he was pushing the business to grow at a rate that’s equal to or above its current U.S. commercial growth rate for the next 18 months. That segment alone has grown by 149% year-over-year.
Investors need to treat that carefully and not treat it as formal guidance as of yet.
To be fair, for a stock trading at over 80 times non-GAAP forward earnings, even exceptional growth is unlikely to dispel valuation concerns.
Nevertheless, if demand for sovereign AI continues to accelerate, Palantir might have more room to grow beyond its already-premium valuation. On the flip side, if growth slows sharply, Karp’s own language will naturally raise the market’s expectations and leave far less room for execution errors.
Related: JPMorgan makes surprising S&P 500 call after inflation shock
