As the race to build more powerful artificial intelligence models intensifies, Palantir CEO Alex Karp believes the industry’s next defining conflict will not be about model intelligence. Instead, he argues that control over proprietary data will determine which companies emerge as long-term winners in the AI economy.
Palantir Chief Executive Officer Alex Karp has delivered a stark warning to businesses embracing artificial intelligence, arguing that the industry’s next major battleground will revolve around data ownership rather than the intelligence of AI models themselves.
Speaking during a recent CNBC interview, Karp said many organisations have become overly focused on gaining access to the most advanced large language models while paying insufficient attention to protecting one of their most valuable assets, their proprietary data.
According to Karp, businesses risk handing over years of institutional knowledge and competitive advantage if they allow external AI providers unrestricted access to their internal information.
«”These models have been completely, irresponsibly, oversold,” Karp said, criticising what he described as excessive hype surrounding today’s frontier AI systems.»
His comments come as enterprises worldwide continue investing billions of dollars in artificial intelligence following the explosive adoption of generative AI technologies.
While companies such as OpenAI, Anthropic, Google and xAI continue competing to build increasingly capable models, Karp believes many executives are beginning to ask a different question: who ultimately controls the data powering those systems?
The Palantir chief argues that proprietary business data is becoming a strategic asset whose value may eventually exceed that of the AI models themselves. If organisations surrender control of that information, they may unintentionally give away insights that took years or even decades to develop.
Those concerns form the foundation of Palantir’s recently published manifesto on AI sovereignty, which encourages organisations to retain ownership of both their data and AI capabilities instead of becoming dependent on external providers. The manifesto also criticises what Palantir calls “tokenmaxxing,” a growing tendency among companies to spend heavily on AI tokens and model usage without generating meaningful business value.
According to the company, large AI budgets alone should not be mistaken for genuine technological progress. Karp suggested many executives are beginning to recognise that problem. Rather than chasing increasingly expensive AI models, businesses are becoming more interested in practical systems capable of improving productivity while protecting sensitive corporate information.
He also expressed concern about enterprise intellectual property flowing into external AI platforms without sufficient clarity regarding how that information may be used in the future. Those remarks reflect a broader debate emerging across the artificial intelligence industry.
See Also: Meta Upgrades AI Glasses with Panoramic Instagram Stories and Smarter Editing Tools
For much of the past two years, attention has centred on which company possesses the smartest chatbot or the most capable language model. Increasingly, however, governments and enterprises are shifting their focus toward questions of security, governance, privacy and digital sovereignty.
European governments have already begun investing in domestic AI models to reduce dependence on foreign technology providers. Large enterprises are similarly exploring ways to deploy AI while ensuring sensitive customer information and proprietary business knowledge remain under their own control.
Palantir has positioned itself at the centre of that conversation. Founded in 2003, the company built its reputation developing software for governments, defence agencies and large enterprises handling highly sensitive information. Its Artificial Intelligence Platform integrates large language models into secure environments while allowing customers to maintain control over their own data.
Karp believes that approach will become increasingly important as artificial intelligence moves from experimentation to enterprise-wide deployment. He argued that organisations should not simply purchase access to AI models but should instead develop long-term strategies that protect institutional knowledge while creating sustainable competitive advantages.
His criticism also reflects growing frustration among some enterprise customers regarding the cost of deploying frontier AI systems. Several business leaders have begun questioning whether expensive AI subscriptions and usage-based pricing models consistently deliver returns that justify their cost.
Despite his criticism of the industry’s direction, Karp acknowledged that artificial intelligence will remain one of the defining technologies of the coming decade. His message, however, is that companies should focus less on chasing every new AI model and more on safeguarding the information that makes those models valuable in the first place.
Industry analysts believe the discussion around data ownership will become increasingly significant as AI adoption accelerates across healthcare, finance, manufacturing, government and defence. Businesses that retain control over their proprietary information may enjoy greater flexibility to adopt new AI technologies without sacrificing their competitive edge.
Those that fail to protect their data could find themselves increasingly dependent on external providers for technologies built using their own institutional knowledge.
Artificial intelligence continues advancing at remarkable speed, but Karp argues the companies that ultimately succeed will not necessarily be those with access to the smartest models.
They will be the organisations that understand the enduring value of their data and ensure they remain its owner, not merely its supplier.
