PALANTIR: War, Surveillance, Artificial Intelligence

Part 1: The Heretic of Silicon Valley

Chapter 1. The PayPal Mafia and 9/11

Attorney Kyungjin Kim

A. 9/11 and the Lessons of Intelligence Failure

(1) The Tragedy of Disconnected Information (Silos) and the Limits of the CIA

At 8:46 AM on September 11, 2001, American Airlines Flight 11 struck the North Tower of the World Trade Center. Seventeen minutes later, United Airlines Flight 175 tore through the South Tower. In the underground situation room at CIA headquarters in Langley, Virginia, analysts were staring at their screens. None of them had anticipated this. Or, to be precise, it was something they could have anticipated but failed to.

Intelligence agencies began combing through their own files. The CIA knew that two al-Qaeda operatives, Nawaf al-Hazmi and Khalid al-Mihdhar, had entered the United States. They had first identified these two men while monitoring an al-Qaeda meeting in Kuala Lumpur, Malaysia, in January 2000. But the CIA did not pass this information to the FBI. The FBI handled domestic matters; the CIA dealt with foreign intelligence. An invisible wall stood between the two agencies.

Meanwhile, an agent at an FBI field office had submitted a report that summer noting that Middle Eastern men were receiving training at flight schools across the United States. He warned that this could be connected to a terrorist plot. But the report was buried in someone's desk drawer at FBI headquarters. The laptop computer of Zacarias Moussaoui, who had been arrested in Minnesota, contained a flight simulation program. FBI field agents applied for a warrant to search his computer, but headquarters rejected the request. The reason given was that the wording of the warrant application did not meet legal standards.

The NSA, the National Security Agency, had been intercepting communications throughout that summer indicating that "a massive attack was imminent." They reported this information to the President. The title of the Presidential Daily Brief dated August 6 was "Bin Laden Determined to Strike in US." However, this imminent-attack briefing contained no specific dates, locations, or methods. The NSA considered its mission to be collecting intelligence, not analyzing or connecting it.

It was not that there was no intelligence. Intelligence was overflowing. The CIA, FBI, NSA, State Department, Immigration and Naturalization Service , all of these agencies held a piece of the puzzle. The fact that the terrorists' passports were forged, that they had attended flight schools, that they were connected to al-Qaeda, that they were inside the United States. But these pieces were scattered across different rooms.

Each agency had its own database, used its own classification system, and followed its own security regulations. For a CIA analyst to access FBI investigative records, a complex administrative process was required. In most cases, such attempts were never even made.

This is called an "information silo." A silo is a cylindrical storage structure used on farms to store grain. There is a wall between the silo holding wheat and the silo holding corn. The grain on one side does not flow to the other. That was the state of the American intelligence system before 9/11. Intelligence was collected but did not flow. It was not shared. It was not connected.

Two years later, the 9/11 Commission reached this conclusion: "Our principal failure was a failure to connect the dots." The commission pointed out that this failure was not due to individual incompetence but to the structure of the system. The intelligence system, designed during the Cold War, had been built to face a single adversary: the Soviet Union. But the new enemy, al-Qaeda, was different. They crossed borders, moved like a network, and followed no rules. A bureaucratic structure divided into departmental silos could not counter such an enemy.

After the commission's report was released, the U.S. Congress established the Office of the Director of National Intelligence (ODNI). The purpose was to tear down the walls between intelligence agencies. But changing the organizational chart alone was not enough.

The real problem was the data. How to connect hundreds of millions of documents, videos, communications records, and financial transaction records stored in different formats across dozens of agencies. It was impossible for the human eye. Computers were needed. But the computer systems the government had at the time had been designed in the 1980s. They ran different operating systems, were written in different programming languages, and were incompatible with one another.

One man read this report. He was not an intelligence officer. He was not a soldier, nor a politician. He was an entrepreneur who ran an online payments company in Silicon Valley. His name was Peter Thiel.

(2) Peter Thiel's Vision: "Technology that preserves freedom while ensuring security"

One day in 2002, Peter Thiel was sitting in a cafe in Palo Alto, drinking coffee and reading the newspaper. The interim report of the 9/11 Commission was on the front page. "Failed to connect the dots." Thiel could not take his eyes off that sentence. Because he had already solved that exact problem before.

Thiel had just sold PayPal to eBay for $1.5 billion. PayPal was an online payment service. Millions of people sent and received money over the internet every day. The problem was that a significant portion of those transactions were fraudulent. The Russian mafia, Nigerian scam rings, and credit card thieves were using PayPal to siphon off money. At one point, more than 1% of all PayPal transactions were fraudulent. It was a level that could have destroyed the company.

Thiel and PayPal's engineers developed a system called "IGOR" to solve this problem. IGOR analyzed millions of transaction records to find suspicious patterns. For example, if multiple transactions occurred from a single account in a short period of time, it raised a flag. If multiple different accounts were accessed from the same IP address, it sent an alert. It combined dozens of variables , transaction amounts, time zones, locations, card types, shipping addresses , to assign a fraud probability score. However, IGOR did not make the final decision. The last judgment was left to humans. The system filtered out suspicious transactions, and analysts verified them.

Thiel thought about it. Terrorists are no different from fraudsters. They also wire money. They buy plane tickets. They rent cars. They lease apartments. They make phone calls. They send emails. They cross borders. All of these actions leave digital traces. If the government could connect these traces, couldn't it prevent terrorism before it happened?

Thiel had another concern. He was a libertarian. He detested the government surveilling citizens' private lives. After 9/11, American society was gripped by fear. The Patriot Act was passed. The government gained the ability to collect call records without a warrant. The argument that "some freedom must be sacrificed for security" was gaining traction. Thiel believed this was a false dichotomy.

"If technology works properly, you can maintain security without giving up freedom."

Thiel said this in a Forbes interview. Instead of the government indiscriminately collecting everyone's data, the idea was to selectively analyze only data with grounds for suspicion. And to log who viewed what data and when. If an analyst tried to access information beyond their authority, the system would block it. Access records could later be audited. This way, terrorists could be found while still protecting ordinary citizens' privacy.

Thiel began assembling people to bring this idea to life. The first person he recruited was PayPal engineer Nathan Gettings. Gettings was the key architect of the IGOR system. The second and third were undergraduates at Stanford University: Joe Lonsdale and Stephen Cohen. Both were studying computer science and had previously interned during Thiel's PayPal days. They began developing a prototype in 2004.

Thiel needed one more thing: a CEO to lead the company. Thiel was an excellent investor, but he lacked confidence as a manager. He had no talent for managing complex organizations, negotiating with government officials, or preaching a vision to employees. So he thought of a friend from his Stanford Law School days. Alex Karp.

Karp was a strange choice. He was not an engineer. He could not program. He had no experience running a startup. He was a philosophy PhD. He had received his degree from the University of Frankfurt in Germany, in neoclassical social theory. His doctoral thesis was on "aggression and political identity." At the time, he was running a small hedge fund in London, managing the assets of wealthy families.

The reason Thiel chose Karp was precisely because of his background. A company that works with the government, a company that handles classified data, needed legal and ethical sensibility. Karp had a law degree and philosophical training. He was the kind of person who instinctively asked, "What happens if this technology is misused?" Thiel needed someone like that.

On May 6, 2003, a corporation named "Palantir Technologies" was registered in Delaware. The company's mission was "developing software to support counterterrorism operations." Thiel invested $30 million in initial funding. Karp became CEO. They set up shop in a small office in Palo Alto and began writing code.

But other venture investors in Silicon Valley showed no interest. Investing in a startup that worked with the government was considered insane. Government contracts were slow, complex, and low-margin. Michael Moritz of Sequoia Capital spent the entire pitch meeting doodling in his notebook. An executive at Kleiner Perkins told them, "Your company will definitely fail."

But one place , exactly one place , showed interest. It was In-Q-Tel, the CIA's venture investment arm.

B. The CIA's Venture Investment: The Union with In-Q-Tel

(1) The Meeting of Silicon Valley Engineers and Intelligence Analysts

One day in 2004, Alex Karp was standing in front of an ordinary-looking office building in Arlington, Virginia. He wore a loose jacket and disheveled hair. The people waiting inside were former spies and Defense Department officials in crisply pressed suits. Karp was not nervous. Or perhaps the problem was that he was too relaxed.

In-Q-Tel was a nonprofit venture investment organization established in 1999 by then-CIA Director George Tenet. The "Q" in the name was a witty reference to Q, the technology developer in the James Bond films.

In-Q-Tel's mission was clear: to identify cutting-edge technologies in the private sector and introduce them to intelligence agencies. After the Cold War ended, the CIA had been falling behind technologically. While Google was revolutionizing web search, CIA analysts were still organizing intelligence reports with word processors and Excel sheets.

In-Q-Tel's CEO was Gilman Louie. He was a video game designer. The flight simulation game he created was so sophisticated that it was used for Air Force pilot training. Director Tenet had recruited him precisely for that reason. Louie understood what cutting-edge technology was and knew how to apply it to real-world operations.

Usually, Louie did not attend pitch meetings in person. Projects had to be screened by his staff before reaching his desk. But Palantir's case was different. Peter Thiel was involved. Thiel had already earned his reputation through PayPal, and his money and network could not be ignored. Louie decided to attend the meeting personally.

There was something the Karp and Palantir team did not know. At that time, the CIA was in desperate need of new data analytics technology. The program the CIA primarily used at the time was "Analyst's Notebook," a product from the British company i2. The interface was decent, but it had severe limitations in data processing capability. When handling large volumes of data, the system would slow down or crash. Analysts had many complaints.

Karp began his presentation. Instead of PowerPoint slides, he explained verbally. Like a philosopher, he started by identifying the essence of the problem. The failure of 9/11 was not an absence of intelligence but an absence of connection. A system was needed to bring together data from each agency, find hidden patterns, and present them to analysts. But this system must not become a black box. Analysts must be able to understand why a particular result appeared. And it must log who viewed what data and when.

Louie nodded. This was exactly what he had been wanting to hear.

In-Q-Tel decided to invest approximately $2 million in Palantir. By Silicon Valley standards, it was a small amount. But something was more important than the money. In-Q-Tel's investment stamped Palantir as "a government-validated company." And even more important than that was "opening the door." In-Q-Tel helped Palantir's engineers obtain security clearances. And it placed them before actual CIA analysts.

At first, there were clashes. Silicon Valley engineers and Washington intelligence analysts were people who spoke different languages. The engineers wore shorts and slept on the office floor. They disliked authority, ignored rules, and their creed was "fail fast and fix fast." The analysts, on the other hand, were people who had spent their entire lives within bureaucracy. They valued procedure, respected hierarchy, and a mistake meant the end of a career.

Palantir's first prototype failed miserably in front of CIA analysts. The system was flashy, but it couldn't do what the analysts actually wanted. One analyst asked, "If I press this button, will it tell me where Bin Laden is?" The engineers were left speechless. They understood the technology but did not understand the users.

So Palantir adopted a new approach. It was the concept of the "Forward Deployed Engineer." Engineers left the comfortable offices of California and went into the underground bunkers of Washington and the forward bases of Afghanistan. They sat next to the analysts. They observed how the analysts worked. They asked what was inconvenient. And they fixed the code on the spot.

This was completely different from the approach of traditional defense contractors. Large defense companies like Lockheed Martin or Raytheon would receive a requirements document from the government, spend years developing a product, and then deliver the finished item. If the product didn't work in the field, it took more years to fix. Palantir was different. They built prototypes within weeks, tested them in the field, received feedback, fixed them, and redeployed. They built software for the government at Silicon Valley speed.

Slowly, the analysts' attitudes began to change. They realized that with Palantir's software, they could complete analysis that previously took months in just hours. They identified patterns in where improvised explosive devices (IEDs) were being planted in Iraq. They tracked the flow of Taliban funding in Afghanistan. They visualized the networks of drug cartels. Word began to spread among analysts: "Try Palantir. It's different."

(2) Establishing the Identity of "Google for Government"

In 2008, Palantir released its first official product, "Gotham." The name was taken from Batman's "Gotham City" , the image of a hero fighting crime in the darkness. Gotham was a platform for large-scale data analysis, integration, and visualization. It was designed for intelligence agencies and military operations.

The core of Gotham was search. But it was a different kind of search from Google. Google searches the open web. It quickly finds information that anyone can see. Gotham searched locked data. Classified documents, intelligence reports, wiretap records, financial transaction records, airline reservation records, immigration records. It allowed information scattered across different databases to be viewed on a single screen.

Here is an example. An analyst searches for the name of a specific terror suspect. Gotham pulls up all information associated with that name. The phone numbers that person used. The people called from those phone numbers. The bank accounts of those people. The flow of money in and out of those accounts. The airline tickets that person booked. The people on the same flights. All of this information is visualized as a graph. Dots are connected by lines. The analyst can see the entire shape of the network at a glance.

That is why people called Palantir "Google for Government." The nickname was apt yet incomplete. Google only finds information. Judgment is left to the user. Palantir was the same. But there was one critical difference. On Google, anyone can search for anything. On Palantir, they could not.

Strict access controls were built into the Palantir system. Every user had a clearance level. One analyst might only have access to Database A; another might only have access to Database B. If someone tried to view information they were not authorized to see, the system blocked them. And all activity was logged. Who opened what file and when, what search terms were entered, what results were exported , all of it was recorded. Auditors could later review these logs.

This was the meaning of Peter Thiel's "security without surrendering freedom." Not indiscriminate surveillance, but controlled analysis. Of course, this system was not perfect. Access permissions had to be configured by someone, and that someone could make mistakes or abuse their authority. But at least it was better than the previous system. Before, analysts could rummage through file cabinets as they pleased, and no records were left behind.

By 2010, 70% of Palantir's clients were government agencies: the CIA, FBI, NSA, Marine Corps, Air Force, Special Operations Command, National Center for Missing & Exploited Children. The remaining 30% were private financial institutions. Large banks like JP Morgan began using Palantir for fraud detection. The technology that started at PayPal had expanded to Wall Street.

Mysterious rumors circulated around Palantir. There was a rumor that Palantir had contributed to the 2011 operation that killed Osama bin Laden. The story was that when U.S. Navy SEAL Team 6 raided the compound in Abbottabad, Pakistan, and killed bin Laden, Palantir's software had been used to track him. Palantir neither denied nor confirmed the rumor. The silence fostered mystique. The mystique brought in new clients.

Palantir defined its identity not as a mere IT service provider but as "a critical infrastructure supplier defending democracy." Alex Karp said in an interview: "We believe software can change the nature of warfare." It was not an exaggeration. In Iraq and Afghanistan, U.S. forces used Palantir to predict insurgent movement routes, analyze IED placement patterns, and track enemy funding flows. Faster intelligence means faster decisions, and faster decisions mean faster operations. On the battlefield, speed was life.

The software created by Silicon Valley's heretics was becoming the nervous system of the world's most powerful military.

C. The Lord of the Rings and Palantir

(1) The Meaning of the Mission: "Seeing Stones"

Peter Thiel was a passionate fan of J.R.R. Tolkien. He had read The Lord of the Rings multiple times since his undergraduate days at Stanford. When naming the company, he recalled a mystical object from the novel. The "Palantir" , plural "Palantiri."

Palantir, in Quenya, the Elvish language Tolkien created, means "that which sees from afar." "Palan" means "far," and "tir" means "to watch." In the novel, the Palantiri are indestructible crystal spheres. Those who possess one can see events happening in distant places, transcending space and time. They can glimpse events of the past. They can communicate with those who hold other Palantiri.

According to Tolkien's mythology, the Palantiri were made by the Noldor Elves of Valinor , perhaps the work of the greatest craftsman, Feanor himself. Eight Palantiri existed in total, and seven were in Middle-earth. The Faithful Numenoreans placed them throughout the kingdoms of Gondor and Arnor. These stones formed a network that surveilled the entire realm. The king could know of distant rebellions in advance and detect intruders at the borders.

Thiel believed this name perfectly symbolized the company's mission. To see through the vast data scattered across the world, detect hidden threats within it, and protect the nation. The tragedy of 9/11 happened precisely because of a failure to "see far." Intelligence agencies could not connect fragmented data and failed to predict the coming attack. Palantir's software aspired to become exactly that "far-seeing eye."

But Tolkien's story also contained a warning. In The Lord of the Rings, the Palantir was a double-edged sword. Saruman, the White Wizard, tried to use the Palantir to monitor the enemy's movements. But he gazed too deeply. The Dark Lord Sauron held another Palantir, and Saruman's mind was dominated by Sauron. He was corrupted and became Sauron's servant.

Denethor, the Steward of Gondor, also used a Palantir. He sought to understand the enemy's forces to defend his kingdom. But Sauron used the Palantir to show Denethor visions of despair. Denethor saw the overwhelming enemy army and lost all hope. In the end, he threw himself into the flames and died.

Tolkien scholar Jane Chance Nitzsche analyzed it this way: Saruman fell because he chose "mere knowledge" and abandoned "true wisdom." The Palantir gave information, but it did not give the wisdom to interpret that information correctly. Instead, it caused the loss of judgment in a flood of information.

Thiel and Karp were fully aware of this literary warning. They knew that the technology they were building could become like Saruman's Palantir. Powerful surveillance capability, when misused, becomes a tool of totalitarianism. That is why they placed access controls and audit logs at the core of the system. The principle was: "We can see everything, but we do not look at what should not be seen."

Palantir's offices also followed Tolkien's world. The Palo Alto headquarters was called "The Shire," the homeland of the Hobbits. The McLean, Virginia, office was "Rivendell," the Elven refuge. The Washington, D.C., office was "Minas Tirith," the capital of the Kingdom of Gondor. Every conference room bore a Tolkien place name. New employees received The Lord of the Rings as required reading.

This was not mere branding. It was Palantir's identity itself. They likened themselves to "the alliance of free peoples defending Middle-earth." The enemy was Sauron , terrorists and totalitarian states. Their weapons were not swords and bows but code and algorithms. And the Palantir they built was the "seeing stone" that protected Western civilization.

Critics saw it differently. They warned that Palantir could itself become the Eye of Sauron. A totalitarian tool for the state to surveil its citizens. The irony of destroying freedom in the name of protecting it. The duality embedded in the name "Palantir" became a debate that ran through the company's entire history.

(2) Seventeen Years of Silence: The Journey from Unicorn to Public Listing

On Wednesday, September 30, 2020, the opening bell rang at the New York Stock Exchange (NYSE). The ticker "PLTR" appeared on the screen for the first time. It was the moment Palantir Technologies' stock began trading on the public market. Starting from a reference price of $7.25, it opened at $10 and closed at $9.50. The market capitalization was approximately $16.5 billion , $21.7 billion on a fully diluted basis.

Palantir was founded in 2003. It took a full 17 years to go public. Considering that Silicon Valley startups typically go public or get acquired within 5 to 10 years, this was an exceptionally long time.

Why did it remain private for so long?

The first reason was the nature of its clients. Palantir's main clients were the CIA, NSA, FBI, and the Department of Defense. Spies and special forces operators. They intensely disliked having the software they used made public. Becoming a public company means disclosing financial statements. Revenue breakdowns must be revealed. Which government agencies paid how much could be exposed. The clients did not want that.

The second reason was Alex Karp's philosophy. Karp despised Wall Street's pressure for short-term results. He refused to change strategy to meet analysts' quarterly expectations. He thought in terms of 10 and 20 years. The goal was to build the technological infrastructure that defends Western civilization. Such a long-term project was difficult to reconcile with quarterly earnings reports.

In 2013, Karp said in a media interview: "Going public would make it very difficult to run a company like ours." That December, Palantir raised $450 million from private investors. The company was valued at $9 billion. Forbes called Palantir "one of the most valuable private technology companies in Silicon Valley."

In 2014, it raised an additional $50 million, and the valuation rose to $15 billion. In 2015, it secured $880 million in funding and reached a $20 billion valuation. The term "unicorn" was no longer sufficient. Palantir had become a "decacorn" , a company worth ten times $1 billion.

Behind the flashy numbers, there was pain. Palantir recorded losses of hundreds of millions of dollars every year. A net loss of $598 million in 2018. A net loss of $588 million in 2019. It was not making a lot of money , it was spending a lot of money. The problem was the business model. For a long time, Palantir operated like a consulting firm. It dispatched engineers to client sites to custom-install software and modify it to meet each client's specific needs. This was an expensive approach. Acquiring a single client required millions of dollars and months of time.

A turning point came in 2016. Palantir launched a new platform called "Foundry." Unlike Gotham, Foundry was designed for private-sector companies. It was a tool that allowed companies in manufacturing, finance, healthcare, and energy to integrate and analyze their own data. Foundry was more standardized than before. Customization was reduced, and deployment time was shortened from months to days. The cost structure began to improve.

In 2018, Morgan Stanley valued Palantir at $6 billion , a sharp drop from $20 billion in 2015. Investors' patience was running thin. How long could one keep betting on a company that had been posting losses for 17 years? Inside Palantir, discussions about going public began.

Then in 2020, the COVID-19 pandemic hit. Ironically, the pandemic was an opportunity for Palantir. The British National Health Service (NHS) used Palantir's Foundry for vaccine distribution and bed management. The U.S. Department of Health and Human Services used Palantir software called "Tiberius" to manage vaccine allocation. Suddenly, Palantir had become a core supplier of national health infrastructure. Revenue surged.

In September 2020, Palantir finally entered the public market. But instead of a traditional IPO, it chose a "Direct Listing." In an IPO, the company issues new shares, investment banks set the price, and the company pays fees to the banks. In a direct listing, existing shareholders sell their shares directly on the market. No new capital is raised, but there are no commissions to pay to Wall Street intermediaries.

Karp chose this method deliberately. He despised Wall Street conventions. The day before the listing, he released a video message filmed in the woods of New Hampshire. He was wearing cross-country ski gear and a helmet. Removing his helmet, he said: "We believe our software makes the world safer. If you don't agree with that belief, don't buy our stock."

On the first day of trading, Karp said in a CNBC interview: "We have built the foundation on which the company carries substantial value. Entering the public market will help our relationships with clients and accelerate growth."

Five years later, by September 2025, Palantir's stock price had risen 1,700% from the first day of trading. The market capitalization surpassed $430 billion. Quarterly revenue broke through $1 billion for the first time. The number of clients grew from 125 in 2020 to 849. Headcount also increased by more than 1,500.

Seventeen years of silence and patience were rewarded. Peter Thiel's original vision , "technology that preserves freedom while ensuring security" , was valued at hundreds of billions of dollars on Wall Street. But whether the future Palantir reveals is truly bright, or whether a dark shadow looms over it as Tolkien warned, that judgment remains in the reader's hands.