Palantir Technologies (PLTR) FY2025 results: revenue $4.5 billion, up 56.2% on the year

Palantir builds software that lets an organisation integrate its data, decisions and operations. It was founded in 2003 and started by building software for the United States intelligence community for counterterrorism investigations and operations, later extending to commercial enterprises. There are four principal platforms: Palantir Gotham, Palantir Foundry, Palantir Apollo and the Artificial Intelligence Platform. It had 954 customers as of 31 December 2025. Of the revenue it earned in 2025, 54% came from the government segment and 46% from the commercial segment, with 74% from customers in the United States and 26% from abroad.

The four platforms

Foundry is the data operations platform, covering data management, logic authoring, systemic mapping, analytics and workflow development. Individual users integrate and analyse the data they need in one place, connecting models across workflows, teams and organisational units, and the interfaces let a user build data pipelines, run high-scale analytics and build operational applications. The filing states that all of the company's commercial customers now use it, as do many government customers.

Gotham integrates data from across domains and sensors in near real-time so users can see, understand and act in the modern battlespace, from operations centres to the tactical edge. It also handles the hand-off between analysts and operational users, so operators can plan and execute mission plans while holding a complete operational picture across echelons. It is used broadly across government functions.

The Artificial Intelligence Platform, which the company began deploying in 2023, gives unified access to open-source, self-hosted and commercially available large language models, turning structured and unstructured data into objects a model can read and turning an organisation's actions and processes into tools for humans and model-driven agents. It carries interfaces for decision making, feedback and hand-off between AI agents and human operators, with security and audit controls giving granular control over model usage and human review checkpoints inside the workflows. It is bundled with Foundry, Gotham and Apollo.

Apollo is the continuous delivery platform, a cloud-agnostic single control layer that coordinates delivery of new features, security updates and platform configurations. It was originally built to deliver Palantir's own software wherever a customer is, in the cloud, on-premises or in more rugged environments, and customers can now use it to deploy their own software in almost any environment.

The Ontology

The Palantir Ontology is the systematic mapping of data to meaningful context, and the company describes it as the heart of the platforms. It integrates the elements of a decision, being the data, the logic and the actions, into one representation of the organisation, so users can build interconnected workflows. The filing says the problem it is aimed at is that data projects often fail because data pipelines sit disconnected from the analytical and operational work, and that the Ontology provides a common system for data, analytics and operations. Access is controlled finely, with permissions propagating from source data through to shared analyses.

Who uses it

The software reaches users across business functions and levels, and the filing names utility operations analysts, automotive manufacturing workers, oil and gas technicians and operators, and pharmaceutical researchers in the United States; supply-chain managers in South Korea; public health administrators in the United Kingdom and the United States; and special forces personnel and military officials in the United States and abroad.

How it wins the work

The customer acquisition strategy targets large-scale, hard-to-execute opportunities at large government and commercial institutions. The company's argument for that choice is set out plainly in the filing: high installation costs, high failure risk, complex data environments and long sales cycles raise the barriers to entry, and it states that the larger and more technologically demanding the problem, the more likely it is to succeed. It gives the example of one U.S. military department that spent more than $1 billion building an enterprise resource planning system from scratch, which was never delivered and was terminated, and says such failures have left software buyers and vendors risk-averse.

Against that, speed of deployment is what widens the reach. AIP bootcamps deliver real workflows on actual customer data in days, and a Developer Tier gives limited access to Foundry and AIP in the United States and select countries without large upfront enterprise costs. Teams embed directly with customers across industries.

From Palantir's annual report on Form 10-K for the year ended 31 December 2025. The filing

Written from the 10-K filed 17 February 2026: read the full 10-K.