Palantir has raised its annual revenue forecast after another quarter of rapid US growth, although its momentum is exposing a widening contrast between American demand for its data platforms and European resistance around procurement, sovereignty, and public accountability.

The company reported second-quarter revenue of $1.94 billion, an increase of 93 per cent from a year earlier, while adjusted earnings reached $0.41 per share. Palantir now expects full-year revenue of between $8.15 billion and $8.16 billion, almost $500 million above its previous forecast.

US commercial revenue increased by 149 per cent to $764 million, while revenue from the US government rose by 90 per cent to $809 million. Those two markets accounted for most of the quarter’s business, leaving Palantir’s international performance increasingly peripheral to its financial growth.

Yet the European picture remains commercially and politically important because public bodies are deciding how much operational data, analytical capability, and institutional knowledge should be concentrated inside platforms supplied by a US defence technology company.

Growth is concentrating in the United States

Palantir’s expansion reflects demand for software that connects information from multiple systems and turns it into operational workflows. Its platforms are used in defence, intelligence, healthcare, manufacturing, and other sectors where organisations want AI to work with controlled internal data rather than through a standalone public chatbot.

That proposition has gained traction in the US, where government spending, defence programmes, and commercial AI projects are supporting exceptional growth. Palantir closed $3.37 billion in contracts during the quarter, while adjusted free cash flow exceeded $1.2 billion.

The company describes its technology as a way for customers to retain control over models and data, a position that overlaps with European concerns about digital sovereignty. However, sovereignty depends on more than where information is stored or whether a customer can select an AI model.

Public organisations must also consider who controls the underlying software, how dependent operations become on specialist implementation, whether systems can be transferred to another provider, and what legal or political exposure follows from relying on a supplier headquartered in another jurisdiction.

Those questions become more difficult as a platform spreads across an organisation. Data integration can produce substantial operational value, but it can also make replacement expensive because workflows, permissions, analytical models, and staff practices become tied to the same technical environment.

European contracts face a different test

France’s domestic intelligence agency is preparing to replace Palantir’s tools with technology from French supplier ChapsVision. The transition will take place gradually because the existing platform is embedded in sensitive work, illustrating both the strategic desire for a domestic alternative and the practical difficulty of changing providers.

The French government has described the move as part of an effort to avoid strategic dependency on foreign technology. ChapsVision must still prove that a smaller European supplier can match the reliability, functionality, security, and support demanded by an intelligence agency, so localisation alone will not settle the procurement argument.

In London, Palantir is challenging a decision to block a proposed two-year Metropolitan Police contract worth up to £50 million. The Mayor’s Office for Policing and Crime said the procurement had not adequately demonstrated value for money, while Palantir argues that the decision was unlawful and improperly influenced by views about the company’s values and ethics.

The dispute shows how public technology contracts can become inseparable from the process through which a supplier was selected. Even where software appears operationally capable, an authority must demonstrate fair competition, value for money, legal compliance, and a defensible basis for entrusting sensitive information to the provider.

Palantir has argued that losing the system could impose operational costs on policing, while the mayoral authority maintains that the proposed procurement lacked sufficient competition. A court will consider the legal arguments, but the wider issue extends beyond a disagreement over one contract.

Automating intelligence analysis changes how information is assembled, prioritised, and presented to investigators. Procurement must therefore address access controls, auditability, changes to models or rules, human oversight, errors, bias, data retention, and the ability to challenge an output that influenced an operational decision.

Similar questions arise around Palantir’s work in the NHS, where the Federated Data Platform has attracted scrutiny over data access, claimed operational benefits, and the evidence used to measure its performance. Large public-sector platforms may take years to implement, making break clauses, benchmarking, and independent evaluation important long before a contract approaches renewal.

Palantir’s US financial performance means the company is not dependent on winning every European contest. Indeed, the quarter suggests it can continue growing while international revenue becomes a smaller proportion of its business.

European governments face a less comfortable calculation. Replacing an established supplier may reduce one strategic dependency while introducing cost, delay, and technical risk, whereas retaining it can deepen reliance on software that becomes harder to dislodge with each additional workflow.

The contrast between Palantir’s earnings and its European setbacks reflects two different markets for the same technology. In the US, demand and contract volume are validating the company’s commercial model; in Europe, public buyers are asking whether technical capability is sufficient when sovereignty, competition, procurement integrity, and institutional control are also part of the purchase.