Palantir's Alex Karp says businesses may be paying to train future AI competitors, not just for intelligence. As enterprise AI spending surges, the debate over who captures long-term value is intensifying among tech leaders and investors
When Microsoft CEO Satya Nadella recently questioned the true cost of enterprise AI, he reignited a debate that Palantir CEO Alex Karp has been pushing for months: Are companies paying twice for artificial intelligence-once for the technology, and again by giving up their own proprietary knowledge?
Karp took this argument further during Palantir's latest earnings call, warning that the so-called "second payment" could ultimately erode a company's competitive edge. As businesses feed prompts, corrections, and workflow data into external AI models, they may be transferring valuable operational know-how to outside providers. Over time, this could allow AI vendors to replicate or even outcompete their clients by embedding that expertise into their own systems.
Competitive Risks in AI Adoption
Palantir's leadership argues that the operational knowledge companies share with AI providers-how employees solve problems, make decisions, and optimize processes-can become part of the provider's broader technology stack. This dynamic, they say, creates an economic imbalance: while enterprises pay for AI services and supply valuable context, the AI vendor retains control over the underlying model and accumulates intelligence that could be used to build rival products.
Chief Revenue Officer Ryan Taylor sharpened the warning, suggesting that companies are effectively "paying to give away their most important secrets." If these insights are used to train future models, businesses risk contributing to the commoditization of their own competitive advantages. Karp calls this accumulated edge a company's "alpha," and insists that organizations must ensure AI compounds that value internally, not for a third party.
Palantir's Financial Momentum
Despite these warnings, Palantir's own business is booming. For the second quarter of 2026, the company reported revenue of $1.935 billion, up 93% year over year and 19% sequentially, beating consensus estimates by $130 million. U.S. commercial revenue surged 149% to $764 million, while U.S. government revenue climbed 90% to $809 million. The company's top 20 customers increased their average spend by 67% to $124 million each, reflecting deeper adoption among major clients.
Profitability also improved, with adjusted operating income reaching $1.194 billion and a 62% margin. Adjusted free cash flow was $1.22 billion, representing a 63% margin. Palantir's "Rule of 40" score-a software industry benchmark combining growth and profitability-hit 155, far above the 40% threshold considered strong for the sector. Management raised full-year revenue guidance to $8.154 billion, implying 82% growth, and expects U.S. commercial growth of at least 134%.
Following the earnings release, Palantir's stock jumped nearly 15% in after-hours trading on August 3, closing at $144.45. However, the stock remains down 21% over the past six months and 30% year-to-date, even after a 570% gain over the past three years, according to Seeking Alpha.
Sovereign AI as a Strategic Response
To address the risks of external AI dependence, Palantir is promoting what it calls "sovereign AI." This approach allows customers to retain control over their data, operating logic, security architecture, and model weights, while maintaining the flexibility to switch models as needed. In this framework, the AI model becomes just one component of a broader system, rather than the core asset controlled by an outside provider.
Palantir claims its software, combined with Nvidia infrastructure, enables companies to fine-tune models within their own environments, outperforming so-called "frontier" models on specific business tasks without surrendering proprietary intelligence. Karp has been direct with investors, arguing that as concerns about closed AI systems grow, spending will likely shift toward private deployments and open-weight models that keep critical knowledge in-house.
This debate over AI value capture is not limited to Palantir and Microsoft. As Wall Street increasingly demands clear payback from AI investments, the competitive landscape is shifting. For example, recent analysis shows that investors are rewarding companies with proven AI revenue streams, while those without clear monetization are losing ground.
Balancing Growth and Valuation
While Palantir's growth rates are impressive, its valuation remains a point of debate. The stock trades at more than 80 times non-GAAP forward earnings, a premium that leaves little room for execution errors if growth slows. Management has expressed confidence in sustaining high growth, particularly in U.S. commercial markets, but investors should be cautious about treating these ambitions as formal guidance.
Palantir's strategy hinges on convincing enterprises that retaining control over their AI infrastructure is essential to protecting long-term value. If demand for sovereign AI accelerates, the company could continue to expand beyond its already lofty valuation. But if the market shifts or growth falters, heightened expectations could amplify downside risk.
For investors and business leaders, the evolving debate over AI ownership, data control, and competitive risk is becoming central to technology strategy. As more companies weigh the trade-offs of external AI adoption, the question of who ultimately benefits from enterprise AI spending will remain at the forefront of boardroom and market discussions.
AI adoption in the enterprise sector is accelerating, but the long-term implications for competitive advantage and value capture are still unfolding. Companies must weigh the immediate benefits of AI-driven productivity against the potential risks of transferring proprietary knowledge to external providers. As the technology matures, the balance of power between AI vendors and their clients will likely shape the next phase of digital transformation in the U.S. economy.
