A narrated chronicle.

It was the spring of 2003 in Palo Alto. In the backyard of a Victorian house, five young founders gathered daily with the feeling that the world had changed forever after September 11. Peter Thiel, recently out of PayPal and fascinated by Tolkien, insisted that the key to preventing the next attack was not more cameras or higher walls, but patiently weaving together the web of data that already existed, scattered across incompatible silos.

Around a folding table, with open laptops, cold pizza, and the smell of California pine, Stephen Cohen sketched graphs of nodes and edges on a whiteboard. Alex Karp, a philosopher with a sharp tongue, questioned every concept aloud, while Joe Lonsdale and Nathan Gettings played chess with SQL queries as if they were pawns.

The company’s name came from Thiel almost as a joke: Palantir, the seeing stones of Middle-earth. Everyone laughed and nodded, unaware that the word would one day trade on Wall Street.

The first secret ally

The founders worked behind closed doors. They needed to design software capable of integrating flight records, call transcripts, banking transactions, and intercepted messages, all with an audit level strong enough to push back against the ghost of Big Brother.

Two years later, a call from McLean changed the story: In-Q-Tel, the CIA’s venture-capital arm, offered funding and, more importantly, access to real intelligence analysts. From 2005 onward, Palantir engineers began coding inside closed rooms where phones stayed outside and windows were covered.

The challenge was brutal: classified data could not leave secure networks, but the software had to evolve daily. That pressure gave birth to one of Palantir’s defining disciplines: forward-deployed engineering, developers who spend weeks or months embedded inside the client’s environment.

Gotham, the first great leap

In 2008, that permanent dialogue produced a product with its own name: Gotham. It allowed analysts to trace connections through drag-and-drop interfaces instead of endless scripts. Every click was logged, every piece of information tagged according to secrecy level. What began as a tool to hunt terrorists soon helped solve financial crimes and human-trafficking networks.

The reputation spread through Washington. The FBI, NSA, Department of Defense, and other agencies wanted to test the software built by outsiders who seemed to understand their real problems.

A foot outside the state: Foundry

With public-sector clients expanding, Palantir faced a new question: what about the private sector? Many employees wanted to apply the same technology to ordinary but complex problems: manufacturing delays, pharmaceutical traceability, delivery routes, fleet maintenance, and industrial operations.

Foundry emerged as a command cabin for companies. Airbus used it to optimize fleet maintenance; Merck used it to shorten clinical trials; Ferrari used it to read telemetry in real time. The culture, however, remained similar: whole teams lived inside the client’s plant as if every factory were a new military base.

The lights of Wall Street

Seventeen years after that cold pizza in Palo Alto, Palantir went public through an unusual direct listing. Alex Karp spoke from Colorado, reminding investors that the company had taken nearly as long to go public as Apple took to launch the iPhone.

The market was skeptical at first, but the stock gained momentum as the company signed contracts tied to measurable savings. By 2024, Palantir announced its first full year of GAAP profitability and a strong cash position with no relevant debt.

Apollo and the AI era

Managing software deployed across secret servers, public clouds, and even submarines required a new layer. Apollo made it possible to update Gotham and Foundry without shutting down the machine, even in disconnected or classified environments.

When large language models began dominating headlines, Palantir responded with its Artificial Intelligence Platform. The promise was not cosmetic chatbots, but the ability for analysts to ask questions in ordinary language and receive SQL, Python, maps, or operational plans while respecting every classification level.

Shadows and dilemmas

Palantir’s rise has not been pure glory. Gender-discrimination lawsuits, work with ICE, and accusations of algorithmic opacity have kept the company under scrutiny. Karp often answers that neutrality does not exist and that the company chooses the clients it believes in. That statement captures the ethical dilemma: can a piece of software be patriotic while protecting everyone’s freedoms?

If Palantir can sustain the balance between sovereign data infrastructure and democratic oversight, its story may still be beginning. If not, the same doors that opened so quickly may close with equal force.