Palantir Technologies
Last reviewed
Jun 3, 2026
Sources
23 citations
Review status
Source-backed
Revision
v1 · 2,795 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 3, 2026
Sources
23 citations
Review status
Source-backed
Revision
v1 · 2,795 words
Add missing citations, update stale details, or suggest a clearer explanation.
Palantir Technologies Inc. is an American software company that builds platforms for integrating, analyzing, and acting on large and fragmented datasets, with a long history in government and defense work and a fast-growing commercial business. Founded in 2003 by a group that included Peter Thiel and Alex Karp, the company spent its first decade selling almost exclusively to intelligence and defense agencies before expanding into corporate analytics. Since 2023 it has reoriented much of its messaging and product roadmap around enterprise artificial intelligence, packaging access to commercial large language models inside its Artificial Intelligence Platform (AIP). Palantir trades on the Nasdaq under the ticker PLTR and was the single best-performing stock in the S&P 500 during 2025. [1][2][3]
The company is also one of the more polarizing names in technology. Supporters see a rare profitable software firm that takes hard national-security problems seriously; critics see a surveillance contractor whose tools increasingly underpin domestic data aggregation, immigration enforcement, and battlefield targeting. Both readings draw on the same set of facts. [4][5]
Palantir was incorporated on May 6, 2003. The founding group was Peter Thiel, Alex Karp, Stephen Cohen, Joe Lonsdale, and Nathan Gettings. Thiel, who had co-founded and led PayPal, supplied the original idea and most of the early money. PayPal had built statistical fraud-detection systems that flagged suspicious transactions while keeping a human analyst in the loop, and Thiel argued that a similar approach, software that helps a trained person link records and spot patterns without drowning in data, could have helped intelligence agencies connect the dots before the September 11 attacks. Karp, an old Stanford Law School acquaintance of Thiel's with a doctorate in social theory from Frankfurt, was brought in as chief executive, a role he still holds. [4][6]
A prototype built largely by Gettings, Lonsdale, and Cohen became the basis for the company's first product, later named Gotham. Traditional venture firms showed little interest in a startup whose main prospective customer was the U.S. intelligence community, so Thiel's fund bankrolled most of the roughly $30 million it took to get going. The notable outside backer was In-Q-Tel, the venture arm affiliated with the CIA, which put in about $2 million around 2005 and, more importantly, gave Palantir access to agency analysts who helped shape the software. That early grounding in classified workflows, fine-grained permissions, and auditable access still shows up in how the products are designed. [4][6]
Palantir describes its software as an operating system for an organization's data. The platforms share a common idea: rather than build one-off dashboards, model the real-world objects a customer cares about (people, equipment, shipments, transactions) as an "ontology," then let users and, more recently, AI agents act on that model. [7][8]
| Product | Launched | Primary market | What it does |
|---|---|---|---|
| Gotham | ~2008 | Defense, intelligence, law enforcement | Integrates classified and open datasets, links entities, visualizes networks, and supports targeting and investigations under strict access controls |
| Foundry | ~2016 | Commercial and civil government | A data-integration and operations platform that builds a shared ontology across siloed corporate systems for supply chain, manufacturing, finance, and health uses |
| Apollo | 2021 (productized) | Internal and customer deployment | Continuous delivery infrastructure that installs and updates the other platforms across cloud, on-premises, and classified or disconnected environments |
| AIP | April 2023 | Both government and commercial | A layer that wires LLMs and other models into Foundry and Gotham data with governance, letting users query data in natural language and build agents and automations |
Gotham is the original product and the one most associated with Palantir's reputation. It pulls together disparate sources into a single graph an analyst can explore, and it has been used for counterterrorism, fraud investigations, and military command. Foundry, launched in the mid-2010s, was the company's bid to escape its dependence on government contracts by selling the same core technology to corporations under a friendlier framing. Apollo, the deployment system, is less visible to outsiders but is what lets Palantir push updates into a CIA network and a commercial cloud tenant from the same pipeline. [7][8]
The Artificial Intelligence Platform, announced in April 2023, is the product that reframed Palantir as an AI company. AIP does not train its own frontier model. Instead it acts as a governed bridge between a customer's data and a menu of third-party models, with support for LLMs and embedding models from providers including OpenAI, Anthropic, Google, Meta, and xAI. The pitch is that an LLM is only useful inside an enterprise if it can safely reach proprietary data and trigger real actions, and that Palantir's permission model and ontology make that possible without leaking sensitive records into a public chatbot. [3][9]
In practice, AIP exposes a set of builder tools. AIP Logic lets developers define LLM-backed functions; what Palantir originally called AIP Agent Studio (later AIP Chatbot Studio) is used to create conversational agents grounded in the ontology; and AIP Evals is a test harness for measuring whether those AI workflows behave correctly before they go into production. The system enforces the same access controls, encryption, and audit logging as the rest of the platform, so an agent acting on a user's behalf can only see and do what that user is allowed to. [3][9]
Palantir's main go-to-market tactic for AIP has been the "bootcamp," a one-to-five-day workshop in which a customer's engineers build a working use case on their own live data alongside Palantir staff. The company credits these short, hands-on sessions with shortening sales cycles and driving much of its commercial growth since 2023. [10]
Government work remains, in Karp's words, "the constant core" of the company. Gotham and, increasingly, AIP are used across U.S. intelligence and defense, and Palantir has become one of the more prominent software vendors in the recent wave of military AI procurement. [2][11]
A few contracts stand out. In March 2024 the U.S. Army selected Palantir to build TITAN (Tactical Intelligence Targeting Access Node), a ground station that fuses data from space and high-altitude sensors and pushes targeting information to soldiers; the award covered prototype production worth roughly $178 million. In May 2024 the Department of Defense signed an initial five-year contract, reported at about $480 million, for Palantir's Maven Smart System, the AI targeting and decision-support software that grew out of Project Maven. Demand pushed that program's ceiling far higher: a 2025 modification added hundreds of millions of dollars and the overall ceiling rose toward $1.3 billion through 2029. In April 2025 NATO agreed to acquire a version of the Maven Smart System for its Allied Command Operations, in a procurement that NATO described as one of the fastest in its history. [11][12][13]
Beyond the U.S., Palantir has done significant work in the United Kingdom, including a 2023 contract to build the NHS Federated Data Platform, and it has provided software to Ukraine's government and military. The company has been open about the nature of this work; Karp has said publicly that Palantir software is, on occasion, used to kill people, and has argued that Western technologists have an obligation to support their countries' defense. [5][14]
The commercial side, anchored by Foundry and AIP, has been the company's fastest-growing segment. Customers span manufacturing, healthcare, energy, financial services, and aerospace, and have included firms such as Airbus, Morgan Stanley, and Ferrari. U.S. commercial revenue has grown far quicker than the rest of the business, more than doubling year over year in several recent quarters, and the U.S. commercial customer count reached 615 by the first quarter of 2026, up 42% from a year earlier. [2][15]
Palantir has also pushed an industrial-policy angle with a program it calls Warp Speed, a manufacturing-focused operating system aimed at companies trying to re-shore production in the United States. Early Warp Speed customers announced in March 2025 included defense and hardware startups such as Epirus, Red Cat, Saildrone, Saronic, Ursa Major, and SNC. [15]
Karp has run the company since its founding and is known for an idiosyncratic public persona, shareholder letters dense with literary and philosophical references, and a willingness to feud with critics. Peter Thiel chairs the board, and co-founder Stephen Cohen is president. The senior team also includes Shyam Sankar as chief technology officer and executive vice president, Ryan Taylor as chief revenue officer and chief legal officer, and David Glazer as chief financial officer. [6][16]
| Role | Person |
|---|---|
| Chief executive officer | Alex Karp |
| Chairman | Peter Thiel |
| President | Stephen Cohen |
| Chief technology officer and EVP | Shyam Sankar |
| Chief revenue officer and chief legal officer | Ryan Taylor |
| Chief financial officer and treasurer | David Glazer |
Palantir went public on September 30, 2020, through a direct listing on the New York Stock Exchange rather than a traditional IPO. In November 2024 it moved its listing to the Nasdaq Global Select Market, keeping the PLTR ticker, a step tied to eligibility for the Nasdaq-100 index; it had already joined the S&P 500 in September 2024. [1][17]
For years the company posted large losses on a GAAP basis, weighed down by stock compensation, which fed a long-running debate about its valuation. That changed as the business scaled. The figures below trace the recent trajectory. [2][18][19]
| Period | Revenue | YoY growth | GAAP net income | Notes |
|---|---|---|---|---|
| FY 2024 | $2.87 billion | 29% | (full-year GAAP profit) | First full year of GAAP profitability |
| FY 2025 | $4.48 billion | 56% | $1.63 billion | Income from operations of $1.41 billion; Q4 Rule of 40 reached 127% |
| Q1 2026 | $1.63 billion | 85% | $871 million | US revenue +104%; adjusted operating margin 60% |
The first quarter of 2026, reported in early May, was the company's strongest on record by several measures. Revenue of about $1.633 billion was up 85% year over year; U.S. revenue grew 104%; U.S. commercial revenue rose 133% to roughly $595 million; and U.S. government revenue grew 84% to about $687 million. GAAP net income was roughly $870 million. Net dollar retention hit 150%, and the company closed total contract value of $2.41 billion, including 206 deals worth at least $1 million each. Palantir held about $8 billion in cash with no debt, and management raised full-year 2026 revenue guidance to between $7.65 billion and $7.66 billion, implying roughly 71% growth, while lifting U.S. commercial guidance above $3.2 billion. [2]
The stock has been one of the market's wildest rides. Palantir was the top-performing stock in the S&P 500 during 2025, at points up more than 140% on the year and outpacing larger AI names like Nvidia, even as short sellers and skeptical analysts repeatedly warned that the valuation had detached from the underlying numbers. The shares are volatile around earnings; the stock fell several percent immediately after the blowout Q1 2026 report, a reminder of how high expectations had become. [18][20]
Privacy and civil-liberties concerns have followed Palantir throughout its history, and they have intensified as its government footprint has grown. [4][5]
The sharpest recent debate concerns U.S. immigration enforcement. In April 2025 Immigration and Customs Enforcement awarded Palantir a contract, initially around $30 million, to upgrade its Investigative Case Management system and build a platform reported under the name ImmigrationOS, intended to streamline enforcement "from identification to removal." The broader sole-source arrangement has since grown well past $100 million. Civil-rights groups including the American Civil Liberties Union and the Electronic Frontier Foundation have argued that Palantir tools let ICE link individual profiles across many datasets, drawing on sources such as Medicaid and other federal records, to plan operations that critics describe as mass surveillance rather than targeted enforcement. In sworn testimony cited by the EFF in 2026, ICE personnel described using a Palantir tool called ELITE to decide where to conduct deportation sweeps. In April 2026, more than two dozen members of Congress, led by Representatives Dan Goldman and Nydia Velázquez and Senator Ron Wyden, demanded answers from ICE and DHS about the use of Palantir technology. [5][21][22]
Palantir has pushed back forcefully. In a January 2026 response to an EFF report, the company said it was not building any "master database" to unify federal records for the surveillance of citizens and emphasized its own human rights policy and contractual safeguards. Critics counter that legal compliance and a published policy are not the same as ethical restraint, and that the company's denials sit uneasily next to the capabilities its software demonstrably provides. [22][23]
Other episodes have drawn scrutiny over the years. Palantir's NHS Federated Data Platform contract in the United Kingdom prompted protests from patient-privacy advocates and a slow, contested rollout. The company took over the Pentagon's Project Maven drone-analysis work after Google employees revolted and Google declined to renew, a decision Palantir leadership has defended as the right thing to do. In 2011 leaked emails revealed that Palantir had been part of a proposal to discredit critics of WikiLeaks, including the journalist Glenn Greenwald; the company apologized and said the employee involved had been placed on leave. In 2017 Palantir settled, for about $1.7 million, U.S. Department of Labor allegations that its hiring process had discriminated against Asian applicants, without admitting wrongdoing. The company has also drawn occasional links to Cambridge Analytica through an employee who worked with the firm, though Palantir denied any corporate partnership. [4][5][14]
It is genuinely hard to place Palantir on a simple scale. By the numbers it is now one of the more impressive software businesses in the market: profitable, growing fast, flush with cash, and riding real demand for ways to put AI to work on messy enterprise and government data. The AIP strategy, selling governed access to other companies' models rather than competing to build the biggest model, has so far looked smart. At the same time, the work that funds much of this, software that helps governments target, track, and remove people, is exactly the kind of capability that makes a lot of people uneasy, and the gap between what Palantir says its tools are for and how its customers actually use them is not always small. The company tends to treat that tension as a feature rather than a problem. Whether that confidence holds up will depend less on the next earnings beat than on how the surveillance questions get answered. [2][5]