Artificial intelligence has become one of the most powerful business tools ever created. Companies are feeding AI systems customer information, engineering documents, financial models, software code, legal research, marketing strategies, and countless other forms of proprietary knowledge in pursuit of greater productivity. Yet a growing number of business leaders are beginning to ask a troubling question: Who ultimately benefits from all of that information?
One of the loudest voices raising that concern is Alex Karp, the co-founder and CEO of Palantir Technologies. Known for his outspoken style and his work with some of the world’s largest corporations and government agencies, Karp has launched an unusually direct challenge to the growing power of the major AI laboratories. In his view, businesses should think carefully before allowing outside AI providers to become deeply embedded in their operations.
As Karp bluntly stated, “At every single enterprise I deal with, these people are livid.” He argued that many companies are paying significant amounts for AI services that “create no value,” while questioning whether the balance of power has shifted too far toward the companies building the underlying AI models.
His frustration is about far more than pricing.
The Fight Over Enterprise Value
Palantir’s recent white paper, “Institutional Sovereignty in the Age of AI,” argues that organizations must begin treating their data as a strategic asset that requires active protection. The paper outlines numerous steps businesses and governments can take to preserve what it calls data sovereignty.
The concern is straightforward. Every interaction with an AI system potentially reveals something about how a business operates. Internal procedures, engineering techniques, purchasing decisions, customer behavior, pricing strategies, software architecture, manufacturing processes, and countless other pieces of information may collectively reveal the competitive advantages that distinguish one company from another.
Microsoft CEO Satya Nadella recently voiced a similar concern from a different perspective.
“If you’re just a consumer of a foundation model, then I’m not sure how you can retain enterprise value, let alone create.”
That question cuts to the heart of the debate. If AI companies increasingly understand how thousands of successful businesses operate, where will future competitive advantage reside?
What Is Data Sovereignty?
Data sovereignty is the principle that an organization should retain meaningful control over its own information, its business processes, and the knowledge generated through its use of artificial intelligence.
Palantir argues that this is no longer simply an information technology issue. It is becoming a strategic business issue.
According to IBM’s 2026 AI Sovereignty study, 91 percent of surveyed executives admitted they do not fully understand their dependencies across AI vendors, models, and infrastructure. Nearly three quarters reported that changing AI providers would be difficult, while 81 percent said a week-long outage at a major vendor would severely disrupt their business.
IBM concluded that organizations with stronger AI control capabilities protect substantially more operating profit from AI-driven disruptions.
These findings suggest that dependence on outside AI providers may introduce risks extending well beyond technology.
The Importance of Zero Data Retention
One of Palantir’s strongest recommendations is adopting Zero Data Retention, commonly called ZDR.
Zero Data Retention means that prompts, responses, and telemetry are processed only long enough to complete a request before being discarded.
According to the Palantir paper, this approach helps ensure that customer data is not stored, is not used to train future models, and cannot later be examined by other individuals because it was never retained in the first place.
The paper argues that this creates a structural safeguard rather than relying solely on contractual promises.
As the report explains, policies can change over time. Data that was never stored cannot later become subject to changing corporate policies, legal discovery, or future business decisions.
In short, the safest information may be the information that never exists outside your own environment.
Protecting More Than Privacy
Zero Data Retention is only one component of a broader strategy.
Palantir argues that organizations should negotiate ZDR agreements with every model provider rather than becoming dependent upon a single AI company. Maintaining multiple providers preserves negotiating leverage while reducing the risk that any one company can dictate future terms.
The report also emphasizes that data sovereignty requires continuous vigilance. Businesses must understand where their information travels, who processes it, and what technical safeguards exist at every stage of the AI workflow.
Other research points in the same direction.
NTT DATA found that more than 95 percent of organizations recognize the importance of private and sovereign AI, yet fewer than one third are making sovereign AI a near-term priority. The company warns that enterprises built around unrestricted data movement may struggle as privacy requirements and jurisdictional controls become increasingly important.
Security experts are also raising alarms.
TechRadar recently warned that organizations are deploying AI faster than they can properly govern it, creating what it called “one of the biggest security risks facing enterprises today.”
The greatest long-term risk is not necessarily that AI providers misuse customer information today. Rather, it is that businesses become so dependent on a handful of powerful AI companies that they gradually surrender control over one of their most valuable assets: institutional knowledge.
Every prompt submitted, every workflow analyzed, every engineering problem solved, and every strategic decision assisted by AI teaches the system something about how successful organizations operate.
Whether today’s providers would ever seek to leverage those insights is almost beside the point. Incentives change. Business models evolve. Contracts are renegotiated. Corporate policies are rewritten. Companies merge, are acquired, or answer to new shareholders with different priorities.
The prudent executives should assume that any information leaving their organization deserves the same protection as their intellectual property. Data sovereignty is not about distrusting every AI provider. It is about ensuring that a company’s competitive advantage remains under its own control, regardless of how the AI industry evolves.
History has repeatedly shown that the organizations controlling the flow of information often become the organizations controlling the market itself. As artificial intelligence becomes the operating system of modern business, companies that fail to protect their institutional knowledge may eventually discover that their most valuable competitive asset was never their data alone. It was the insight hidden within it.
