30 October 2026 · Leandro Oliva, Giulia Campaioli · #DATAGOV #GovernancebyDataInfrastructure #RegulatoryDataInfrastructures #DigitalSovereignty
In May 2026, the US-based information technology multinational Dell announced a partnership bringing Palantir’s software onto its data-center infrastructure. Palantir is a US company whose software connects data from different sources and helps organizations analyze it. Its public-sector role has attracted controversy, including concerns about surveillance, privacy and dependence on a foreign supplier.
These concerns are often debated through legislation and public announcements. But procurement contracts also shape what governments can do. This post argues that procurement—the process through which public bodies select suppliers and purchase goods and services—can commit institutions to contested technologies before their legal and democratic implications are settled.
When software arrives as infrastructure
On 18 May 2026, Dell and Palantir announced a joint system combining Palantir’s data-analysis and AI software with Dell hardware and NVIDIA computing technology. It is designed to run on premises: on infrastructure controlled by the customer, rather than in a provider’s public cloud.
Dell presents this arrangement as a way to improve security, control and compliance. It also makes Palantir available through an established hardware supplier, as part of a wider infrastructure package.
That does not remove legal obligations or prove that oversight will be bypassed. It does, however, raise a concern about how the purchase is framed. A decision presented as an infrastructure upgrade may receive more attention for its price, compatibility and support arrangements than for the public powers it enables.
For a ministry or hospital, those wider questions matter. Which databases will be connected? Whose information will be analysed? Who can challenge a decision informed by the system? What would it cost to change suppliers?
Procurement can address legal and social requirements. The problem arises when these questions are treated as someone else’s responsibility, while the purchase moves ahead as a technical matter.
Germany: buying before deciding
Baden-Württemberg, a federal state in southwestern Germany, illustrates the consequences of this sequencing. According to reporting by the independent journalism outlet Heise, its interior ministry signed a contract with Palantir worth around €25 million in March 2025, before the legal basis for police use had been established. The contract reportedly contained no exit clause. The ministry justified signing by citing a price offer that would otherwise expire.
The distinction matters: purchasing software is not the same as having permission to use it. But spending millions on a system can put pressure on the subsequent debate. Legislators then face a choice already shaped by the government’s financial commitment.
Earlier experience shows why the legal questions cannot be reduced to purchasing requirements. In February 2023, Germany’s Federal Constitutional Court found provisions in German cities of Hesse and Hamburg unconstitutional. These provisions authorized automated police data analysis without sufficient safeguards. Hesse was already using the hessenDATA platform; Hamburg had not yet used its powers.
The court warned that the laws permitted comprehensive profiles of people and groups “with just one click”. Its concern was informational self-determination: people’s right, in principle, to decide how their personal information is disclosed and used.
This was not a blanket ruling against data-analysis software. It concerned the conditions under which police could use it. Technical capability and purchasing approval could not settle that constitutional question.
The NHS: scrutiny also needs time
England’s National Health Service (NHS) offers a different example. In November 2023, NHS England awarded its Federated Data Platform contract to a Palantir-led consortium, with a maximum value of £330 million over seven years. The platform brings together information used by NHS organizations to coordinate services and resources.
Here, opposition from medical organizations, patient groups and privacy campaigners made the supplier’s role a public issue. The case complicates any claim that procurement necessarily hides politics: purchasing decisions can also become sites of public challenge.
Yet scrutiny must connect with contractual deadlines. The public procurement platform Stotles reports that the initial term ends in March 2027, with a decision on extension due in 2026. The independent evaluation commissioned by NHS England is expected to run until 2029.
The full evaluation is therefore expected after the first extension decision. Interim findings may inform that choice, but the mismatch shows why evaluation schedules and opportunities to leave a contract need to be considered together.
Local servers, wider dependencies
Dell’s emphasis on locally controlled infrastructure raises another distinction. Data residency means where data is physically stored. Jurisdiction concerns which laws and authorities can apply to a company or its activities.
Keeping data on local servers can strengthen control over its location. It does not, however, resolve every question about a foreign supplier’s legal obligations or access to data. Those questions depend on the deployment, access arrangements and applicable law.
Nor does owning the hardware eliminate dependence on proprietary software, updates and support. Digital sovereignty—the ability to exercise meaningful control over digital systems and data—requires more than knowing where the servers stand.
Following the contracts
These examples point to different problems: how a purchase is framed, whether contracts precede legal authorization, and whether evidence arrives before renewal deadlines. Together, they show why procurement deserves attention as part of democratic governance, not merely administration.
For researchers, journalists and policymakers, this means examining contracts alongside laws: what they authorize, when they can end, and how they distribute responsibility. For public bodies, it means bringing technical, legal and public-interest scrutiny together before commitments become difficult to reverse.
The question is not simply whether a system works. It is who gets to decide what it should do—and whether that decision remains open when the contract is signed.
Leandro Oliva is a journalist and researcher pursuing an MA in Cultural Data & AI at the University of Amsterdam. Previously an editor at The Wall Street Journal and The Intercept, his work focuses on AI governance, digital infrastructure, and the political economy of technology. He holds a master’s degree from Columbia University.