silicon valley

Techie Software Soldier Spy

Palantir, Big Data’s scariest, most secretive unicorn, is going public. But is its crystal ball just smoke and mirrors?

Photo: Busà Photography/Getty Images
Photo: Busà Photography/Getty Images
Photo: Busà Photography/Getty Images

Back in 2003, John Poindexter got a call from Richard Perle, an old friend from their days serving together in the Reagan administration. Perle, one of the architects of the Iraq War, which started that year, wanted to introduce Poindexter to a couple of Silicon Valley entrepreneurs who were starting a software company. The firm, Palantir Technologies, was hoping to pull together data collected by a wide range of spy agencies — everything from human intelligence and cell-phone calls to travel records and financial transactions — to help identify and stop terrorists planning attacks on the United States.

Poindexter, a retired rear admiral who had been forced to resign as Reagan’s national-security adviser over his role in the Iran-Contra scandal, wasn’t exactly the kind of starry-eyed idealist who usually appeals to Silicon Valley visionaries. Returning to the Pentagon after the 9/11 attacks, he had begun researching ways to develop a data-mining program that was as spooky as its name: Total Information Awareness. His work — dubbed a “super-snoop’s dream” by conservative columnist William Safire — was a precursor to the National Security Agency’s sweeping surveillance programs that were exposed a decade later by Edward Snowden.

Yet Poindexter was precisely the person Peter Thiel and Alex Karp, the co-founders of Palantir, wanted to meet. Their new company was similar in ambition to what Poindexter had tried to create at the Pentagon, and they wanted to pick the brain of the man now widely viewed as the godfather of modern surveillance.

“When I talked to Peter Thiel early on, I was impressed with the design and the ideas they had for the user interface,” Poindexter told me recently. “But I could see they didn’t have — well, as you call it, the back end, to automatically sort through the data and eliminate that tedious task for the users. And my feedback from the people who used it at the time, they were not happy with it at all. It was just much too manual.”

Smoking his pipe, just as he had when he testified to Congress 33 years ago about his role in facilitating covert arms sales to Iran, Poindexter told me he had suggested to Karp and Thiel that they partner with one of the companies that worked on Total Information Awareness. But the two men weren’t interested. “They were a bunch of young, arrogant guys,” Poindexter said, “and they were convinced they could do it all.”

Seventeen years later, Palantir is seeking to cash in on its ability to “do it all.” Over the years, the company has worked with some of the government’s most secretive agencies, including the CIA, the NSA, and the Pentagon’s Special Operations Command. As recently as two years ago, its value was estimated at $20 billion, elevating it to the loftiest heights of the tech “unicorns,” privately held companies valued at more than $1 billion. On September 30, Palantir is scheduled to go public, selling shares in a highly anticipated gambit that could make Karp one of Silicon Valley’s richest CEOs and cement the reputation of Thiel, the first outside investor in Facebook and a co-founder of PayPal, as one of the most visionary tech entrepreneurs of his generation.

Palantir’s public offering is founded on the company’s sales pitch that its software represents the ultimate tool of surveillance. Named after the “Seeing Stones” in The Lord of the Rings, Palantir is designed to ingest the mountains of data collected by soldiers and spies and police — fingerprints, signals intelligence, bank records, tips from confidential informants — and enable users to spot hidden relationships, uncover criminal and terrorist networks, and even anticipate future attacks. Thiel and Karp have effectively positioned Palantir as a pro-military arm of Silicon Valley, a culture dominated by tech gurus who view their work as paving the way for a global utopia. (Palantir declined to comment for this story, citing the mandatory “quiet period” prior to a public listing.)

It’s a strange moment, given the widespread alarm over the ever-expanding reach of technology, for a tech company to be marketing itself as the most powerful weapon in the national-security state’s arsenal — wrapping itself in what one Silicon Valley veteran calls “the mystique of being used to kill people.” But as Palantir seeks to sell its stock on Wall Street, even some of its initial admirers are warning that the company’s software may not live up to its hype. More than a dozen former military and intelligence officials I interviewed — some of whom were instrumental in persuading government agencies to work with Palantir — expressed concerns about the firm’s penchant for exaggeration, its apparent flouting of federal rules designed to ensure fair competition, and its true worth. The company has largely succeeded, they say, not because of its technological wizardry but because its interface is slicker and more user friendly than the alternatives created by defense contractors.

“Where you get into trouble is when the software gets so complicated that you have to send people in to manage it,” said one former CIA official who is complimentary of Palantir. “The moment you introduce an expensive IT engineer into the process, you’ve cut your profits.” Palantir, it turns out, has run headlong into the problem plaguing many tech firms engaged in the quest for total information awareness: Real-world data is often too messy and complex for computers to translate without lots of help from humans.

A “heat map” Palantir developed of attacks on U.S. troops in Afghanistan. The software’s user-friendly interface, military insiders say, is a big part of its success. Photo: DEFENSEUPDATE/YOUTUBE

One of the central claims made about Palantir — its creation myth, in essence — is that its software was somehow instrumental in locating Osama bin Laden. The company, which has posted a news story repeating the rumor on its website, likes to shroud its supposed involvement in an air of mystery. “That’s one of those stories we’re not allowed to comment about,” Karp once said in an interview.

The only known basis for the claim, which has been repeated in dozens of articles, comes from The Finish, Mark Bowden’s book on the 2011 raid that killed bin Laden. Bowden does not actually say Palantir was used in the raid, but he credits the company with perfecting the data collection and analysis that Poindexter had initiated with Total Information Awareness in the aftermath of 9/11. Palantir, Bowden writes, “came up with a program that elegantly accomplished what TIA had set out to do.”

No one I spoke with in either national security or intelligence believes Palantir played any significant role in finding bin Laden. Thiel, according to Poindexter, wasn’t even interested in building on TIA’s work. “His people were telling him they didn’t need it,” Poindexter recalled.

From the start, Palantir has drawn on a circle of loyal insiders to build the company. In the late 1980s, as an undergraduate at Stanford, Thiel founded a conservative student publication called The Stanford Review to wage war on what he saw as the university’s liberal agenda, including “mandatory race and ethnic studies” and “ ‘domestic partner’ status for homosexuals.” (Thiel, who is gay, married his longtime partner in 2017.) The Review served as a breeding ground for Palantir: Over the years, according to an analysis by a Stanford graduate named Andrew Granato, 24 of the company’s employees came from the staff of Thiel’s student publication.

Palantir’s initial technology was likewise adopted from one of Thiel’s other endeavors: PayPal. In 2000, engineers at the online-payment company wanted to use software to help identify fraudulent transactions, but they found that computer algorithms alone couldn’t keep up with how quickly criminals adapted. Their solution was a program called Igor, after a Russian criminal who was taunting PayPal’s fraud department, that flagged suspicious transactions for humans to review.

In 2003, after PayPal was sold, Thiel approached Alex Karp, a former Stanford classmate with a Ph.D. in neoclassical social theory, with a novel idea: Why not apply Igor to track terrorist networks through their financial transactions? At the time, the CIA unit responsible for locating bin Laden had little experience, or even interest, in such an approach. Thiel put in the seed money, and after a few years of pitching investors, Palantir got its first major breakthrough in the national-security world with an estimated $2 million investment from In-Q-Tel, a venture-capital firm set up by the CIA. According to a former intelligence official who was directly involved with that investment, the agency hoped that tapping the tech expertise of Silicon Valley would enable it to integrate widely disparate sources of data regardless of format. “I have mixed feelings about the CIA,” Richard Perle told me, “but their angel investment in Palantir may have been their most inspired move.”

In-Q-Tel’s investment provided Palantir with something even more important than cash: the imprimatur of the CIA. As doors started to open in Washington, Palantir began to attract fans in the secretive communities of intelligence and national security. One former senior intelligence official recalled visiting the company in Menlo Park, California, around 2005. Palantir didn’t even have its own space — it was working out of the offices of a venture capitalist involved in the firm. “We go out back to the carriage house, and there were sleeping bags under the desks,” the former official recalled. “That’s where the engineers who were doing the code were actually living and sleeping.”

But contracts with spy agencies were never going to provide Palantir with enough scale to satisfy investors. The company needed new customers, especially in the lucrative world of defense contracting, and Thiel knew just how to get them. In Zero to One, his 2014 book on entrepreneurship, Thiel notes a critical move in PayPal’s success: In the early days, the company essentially paid people to sign up, handing out $10 to each new customer.

Under federal rules for procurement, which are laid out in a telephone-book-size manual, you can’t pay Pentagon officials to buy your product because that would constitute bribery. And it makes no sense to entice individual soldiers to use your product, because they don’t have the power to make procurement decisions. But that, remarkably, is exactly what Palantir did.

Not long after In-Q-Tel’s investment, the company began providing software and training to members of the armed forces about to deploy to Iraq and Afghanistan. Rather than focusing on lobbying the Pentagon from the outside, Palantir introduced its product from inside the military, creating both an internal demand and a network of pretrained users. “They would basically contact the soldiers and say, ‘Hey, I would like to give you some training on this tool you will get in theater. Would you like to get trained on it?’ ” recalled Heidi Shyu, then the Army’s chief weapons buyer.

Chris Ieva, a Marine infantry officer who was attending the Naval Postgraduate School in 2006, was an early beneficiary of Palantir’s unorthodox marketing technique. The school is located in Monterey, California, just down the road from Silicon Valley, and Palantir had already established a foothold at the institution. Ieva was excited when he was invited to visit the tech start-up, where he saw engineers walking around with T-shirts that read GOOGLE IS OUR BACKUP JOB.

But Palantir wasn’t trying to recruit Ieva as an employee. Instead, he said, he got funding worth about $10,000 to support his graduate work, which paid for a high-end computer and access to critical data. Ieva was also supplied with Palantir’s software, which the school was leasing for $19,000 a year; the company provided an analyst at its own expense to work with students. “In return,” Ieva told me, “I had to publish a thesis, and the findings would sort of go back to them.” By the time he deployed to Afghanistan in 2011, Ieva was a true believer in Palantir. He was not only trained to use the company’s software but given a personal version to take with him.

It didn’t take long for word of Palantir to make its way up the chain of command. Intrigued by the software his troops were using, Major General John Toolan, the commander of coalition forces in southwestern Afghanistan, met with Palantir. At the time, the primary software for integrating data on the battlefield was the Distributed Common Ground System–Army, which was being produced by leading defense contractors at a cost expected to exceed $10 billion. But users of DCGS–A, particularly those with no training in intelligence systems, found it to be clunky and prone to crashing.

In Afghanistan, improvised explosive devices were the leading killer of troops, and Palantir allowed users to quickly track where the attacks were taking place. Toolan, impressed with the technology, secured some modest funding to buy the software, which came with the added benefit of a team of company engineers who embedded with his forces in Afghanistan. Palantir’s “forward-deployed engineers,” as they are called, essentially operated as a mobile sales force, customizing the software to the needs of each client. “They sent young Palantir technicians and engineers and software designers with us so that when we had a problem, we worked on it together,” Toolan, who later served as an adviser to Palantir, told me. “And that was a heck of a lot better support than what was coming out of the Army for DCGS–A, which was nothing.”

Palantir quickly made inroads in Afghanistan with both the Marine Corps and the Army. But back at the Pentagon, its bottom-up salesmanship was attracting high-level scrutiny. Shyu, the Army’s chief weapons buyer, warned that it was illegal for soldiers to accept free training and software from Palantir. She resolved the issue by putting the company on a modest contract to pay for what it had been providing for free. Palantir’s marketing campaign had not only worked, but the Pentagon was essentially picking up the tab for it.

But Palantir didn’t want one-off contracts — it wanted to become the Pentagon’s primary provider of intelligence software on the battlefield. That drew the ire of military officials who were invested in DCGS–A. What was this off-the-shelf software with a slick user interface that was challenging the system they had spent years assembling? At that stage, as critics were pointing out, Palantir couldn’t do the type of data integration across the armed services that the Pentagon needed.

Shyu decided to check things out for herself. In 2013, she visited military bases in Afghanistan, including those at Bagram, Kandahar, and Kabul, and spoke with intelligence analysts and soldiers engaged in military operations on the battlefield. At one point, she even watched analysts using DCGS–A and Palantir side by side.

Both software programs operated as a sort of Google for spies, allowing users to search and sift through intelligence and battlefield data. But the user interface for Palantir was more like the everyday programs soldiers were accustomed to running on their PCs or Macs. After loading Palantir on their laptops, soldiers could quickly scroll and click through to different options, whether they were searching for a Taliban leader’s associates or tracking the financial network of a drug kingpin. Palantir also incorporated appealing visual icons to represent things like brigades and battalions, in contrast to the stodgy military interface of DCGS–A, which relied on less intuitive features.

With Palantir, an analyst could pull up a map of an area soldiers were about to patrol, draw a bubble around it, and see where improvised explosive devices had gone off in past weeks or months. With just a few clicks, Palantir created a heat map of potential danger zones. Analysts could do the same thing with DCGS–A, but it took longer. Another plus for Palantir: It didn’t crash nearly as often. Its software wasn’t necessarily any better at parsing intelligence, but Shyu could see why some soldiers, particularly infantry who didn’t have time to learn a complex program, preferred it. “I walked away convinced that Palantir is much easier to use,” she says.

As Palantir established itself with the Pentagon’s leadership, it began to operate more like the top-tier defense contractors it was hoping to replace, hiring lobbyists and spreading money around Washington. It launched an all-out war over the future of DCGS–A with two big guns on its side: Lieutenant General Michael Flynn, the soon-to-be head of the Defense Intelligence Agency, and Representative Duncan Hunter, a Marine reservist from California who had been elected to Congress in 2008.

At a congressional budget hearing, Hunter berated the Army’s senior leadership for not providing Palantir to the troops, leading to a near shouting match with a prominent general. No one quite understood Hunter’s obsession with Palantir. “I won’t question his passion,” a former senior Army official told me. “But sometimes his passion overtook his veracity.”

Flynn also advocated for Palantir and called for the military to replace DCGS–A. “The Army needs to move to a DCGS 2.0 quickly,” he told the Washington Times in 2016. “Frankly, I would even change the name, because it just has such a bad moniker right now.”

That same year, Palantir went nuclear: It sued the Army, accusing it of improperly excluding the company from the competition for the next stage of DCGS–A. In court, Palantir’s lawyers, from the high-powered firm Boies Schiller Flexner, accused two Army intelligence officials of having a vendetta against the company that resulted in “six years of bias and prejudice and irrational behavior from the Army.” Hunter also worked to scuttle the nomination of one of the officials, Lieutenant General Mary Legere, who had been slated to become the first woman to head the Defense Intelligence Agency.

Several former military officials told me there was no personal bias against Palantir. The issue, they said, was the way the company had ignored the rules by giving away free goods and services in the form of Palantir software and training. “They did that in amounts of over a million dollars, I know for sure,” said a former senior Army official who was involved with the issue. “And that was just one customer at an Army-brigade level.” (The company’s allies insisted that Palantir did not violate Pentagon rules.) What’s more, the former official added, the company refused an offer from the Army to incorporate its user interface into the existing system, citing its proprietary technology. “They wanted DCGS–A to be all Palantir,” the official said.

In November 2016, the court rejected Palantir’s “bad faith” allegations but ruled the Army had erred in not allowing the company to compete for the DCGS–A contract. Several former national-security officials I spoke with — including some who support Palantir’s technology — expressed bewilderment over the company’s bridge-burning approach. Suing your customer is not the way the Raytheons or Lockheed Martins tend to do business, but it’s not unprecedented for Silicon Valley. Elon Musk, the founder of SpaceX, sued the Air Force in his quest to break into the market for military rockets. “The lesson of Palantir,” said one former official, “is that sometimes being a jerk pays off.”

Just days after winning its lawsuit against the Army, Palantir had another stroke of good luck. Donald Trump, who had received more than $1 million in campaign and super-PAC contributions from Thiel, won the presidential election. Thiel served on Trump’s transition team, and Palantir suddenly found itself with direct access to top administration officials.

According to emails that Andrew Granato, the Stanford graduate, received under the Freedom of Information Act, Thiel met in January 2017 with Francis Collins, who was on his way to Trump Tower to ask the president-elect to reappoint him as head of the National Institutes of Health. “I am looking forward to learning more about Palantir’s current areas of interest,” Collins informed Thiel after the meeting. At the same time, he nudged Thiel about his reappointment. “Of course I’m also curious about what the next steps will be after the interview in Trump Tower,” he wrote. “Is there any way to predict the timing of a decision?”

In June 2017, Trump announced that he was reappointing Collins as head of NIH. And in September 2018, NIH awarded Palantir a three-year contract worth $7 million.

More taxpayer money quickly followed. In 2019, Palantir won a ten-year contract for DCGS–A worth as much as $876 million. That same year, the company renewed a multiyear contract with Immigration and Customs Enforcement, where Palantir’s software has been used by the same division that targets the families of immigrants for deportation. In April, without soliciting any bids from competitors, the Department of Health and Human Services awarded Palantir nearly $25 million to track national COVID-19 data. And in May, the Department of Veterans Affairs spent $5 million on Palantir’s software to “track and analyze COVID-19 outbreak areas” and to provide timely data on supply-chain capacity, hospital inventory, and lab diagnostics.

The company has lost two of its biggest supporters in Washington. In January, Duncan Hunter resigned from Congress after pleading guilty to corruption. And Michael Flynn, who was forced to resign as Trump’s national-security adviser, is currently fighting to have his case dismissed after he pleaded guilty to lying to the FBI. But one of Thiel’s closest associates has occupied high-level posts in two of the places that matter most to Palantir. In 2019, Michael Kratsios, Thiel’s former chief of staff, was appointed chief technology officer for the White House, where he served as one of Trump’s top technology advisers. In July, Kratsios — a 33-year-old with an undergraduate degree in political science — became acting chief technologist at the Pentagon, where he replaced Mike Griffin, a former NASA administrator with a Ph.D. in aerospace engineering.

Like all smart defense contractors, Palantir has ensured its future in Washington by spreading money around to Democrats as well as Republicans. But despite all its federal contracts and bipartisan connections, Palantir is far from healthy. After nearly two decades of aggressive marketing, the company is still losing money and burning cash. In 2019, although its revenues grew by 25 percent, Palantir lost nearly $600 million for the second year in a row. In that light, analysts say, its decision to go public amid a historic economic downturn could be driven by investors and longtime employees eager to be paid out. “Insiders may be struggling to sell their equity and options in Palantir in secondary markets,” observes Sanford C. Bernstein & Company, which advises institutional investors. “A public listing would potentially end the frustration [that] has apparently been an issue for Palantir management for at least a few years.”

Palantir has struggled to expand its commercial business, a key pillar of its promise to investors. By Karp’s own account, the company considered Palantir Metropolis, a product designed for financial institutions, to be an “unmitigated failure” and scrapped it when it failed to find a market. A much-vaunted joint venture with Credit Suisse to police the bank’s own employees was also deemed a “complete bust.” In addition, Palantir appears to be out of favor in some intelligence and national-security circles. By 2015, according to BuzzFeed News, the company’s relationship with the NSA had ended, and Karp was telling employees that the CIA was “recalcitrant” and didn’t “like us.” (Both agencies declined to comment.)

Palantir has also had surprisingly limited success in marketing itself to police, a seemingly natural customer for software that can track criminal networks. The company worked pro bono for the New Orleans Police Department for six years on a secretive contract to target alleged criminals, but the project appears to have ended in 2018.

The New York Police Department also ended its contract with Palantir several years ago. Zachary Tumin, who served as the NYPD’s deputy commissioner for strategic initiatives until 2017, was a fan of the technology. He dispatched a member of his analytics staff to help Bronx detectives use Palantir to create a map of gang members and their activities, marrying data from social media and a wide variety of government sources. “Being able to bring all that data together into one place helped us get a very precise idea of who was involved around a particular set of events,” Tumin said.

On April 27, 2016, apparently aided by Palantir, the NYPD and ICE conducted the biggest gang raid in New York history. Hundreds of officers and federal agents swooped down on the Bronx, bursting into homes and waking families at gunpoint. According to emails obtained by the Appeal under FOIA, a Palantir employee assisted ICE’s Violent Gang Unit. “FYI we’re all set here,” the employee wrote ahead of the raid. “Helping some random agents get set up but we should have a good viz on everyone.” Prosecutors indicted 120 people for gang activity, though residents protested that many of those swept up in the raid had no gang connections.

Palantir, Tumin recalled, was a “highly effective” tool, but its help didn’t come cheap. Every time the NYPD asked the company to add more data, the price went up. “The more powerful and interesting it became, the higher the tax on you for using it,” Tumin said. In addition, the software required a lot of customizing to fit the NYPD’s needs. “We couldn’t make it work without those modifications,” Tumin recalled. “It was too kludgy — it was built for a generic something or other.”

The need for customization points to a deeper problem for Palantir. The customization is what clients like, but it’s also what could prevent the company from scaling. All those software engineers sleeping under their desks may have been great in 2005, when the company was flush with venture capital, but employing an army of humans to endlessly tweak the software doesn’t exactly presage huge profits. “I used to have a metric when I was in the government,” said the former senior intelligence official who visited Palantir’s engineers back in their sleeping-bag days. “People would come in and say, ‘We’ve got this fantastic automated translation system,’ or automated anything. I would say, ‘Does this use RFOP?’ And they would say, ‘I don’t know what that is.’ ”

The acronym stood for Rooms Full of People, meaning the army of analysts required to clean up the data and crunch the numbers. How good any given data-mining system is depends in large part on what’s lurking behind the curtain. Is it artificial intelligence parsing large data sets of complex financial transactions to find the next terrorist? Or is it a room full of eager software engineers sleeping on the floor? Palantir portrays its software as like its namesake — a crystal ball you gaze into for answers. The company emphasizes that it has reduced the time needed to get its software up and running, and former officials told me Palantir has made big improvements to its back end over the years. But the truth is that it still appears to take a lot of manual labor to make it work, and there’s nothing magical about that.

That distinction did not matter to the soldiers in Afghanistan who were trying to pinpoint IEDs, but it makes a huge difference to potential investors, because Rooms Full of People are not nearly as profitable as simply installing software and walking away. “Here’s the dirty secret of all of these data-analytics solutions,” a former Pentagon research manager told me. “They all claim to take these disparate data sources and put them together and then discover these amazing correlations between variables. But the problem is that all of these data sets are terrible. They’re dirty.” Many types of information, after all, are gathered and processed by humans. It may be entered inconsistently or provided in wildly different formats or riddled with inaccuracies. It’s messy, like the real world it reflects and records, and it doesn’t always fit into software with any sort of mathematical precision.

When I saw a recent demonstration of Palantir software, it became clear that this dirty secret isn’t very secret. The interface struck me as user friendly, something anyone with basic computer literacy could figure out. Want to know how many aircraft are available for a specific mission and how long it will take them to get to their destination? With a simple query, Palantir can tell you. Then I was shown a data set on military personnel, which had to be “cleaned up” to make it usable on Palantir. It wasn’t only a magic code doing the cleanup; it was human beings — and even locating someone who could explain what needed to be done had proven time consuming. “It took many calls to find a subject-matter expert,” one person involved told me.

It sounded a lot like Rooms Full of People.

On September 9, Alex Karp appeared on an investor webcast dressed in bright sports gear and hiking up a trail on roller skis. Often described as “eccentric” or a “deviant philosopher,” he stopped and faced the camera, his unruly curls pointing in different directions, and began to talk about Palantir’s tremendous growth. Some 17 years after Karp and Thiel met with John Poindexter, full of confidence and short on engineering, the company was finally set to go public.

Karp blames the darlings of Silicon Valley, not Palantir, for violating people’s privacy. It’s companies like Facebook and Google, he argues, that are selling their users’ data, while Palantir targets terrorists and criminals. “The engineering elite of Silicon Valley may know more than most about building software,” he observed in the company’s filing to go public. “But they do not know more about how society should be organized or what justice requires.” (His argument ignores the fact that Palantir has been used to analyze data from social media, including Facebook posts.)

Poindexter wrote to me shortly after we spoke about his meeting with Karp and Thiel back in 2003. He had seen a recent article about Palantir, he said, and he was shocked at Karp’s transformation. “Karp was clean shaven and had a conservative, traditional-length haircut,” Poindexter told me. “I have no idea why he changed his image. I would not have recognized him from current photos.”

With Karp, as with Palantir, it’s often hard to know what is real and what is mythmaking. It’s often repeated in articles, for example, that Karp studied in Germany under Jürgen Habermas, perhaps the most influential living philosopher. “The most important thing I learned from him is I couldn’t be him, and I didn’t want to be him,” Karp confided on a recent podcast with a sort of knowing intimacy. In fact, as Moira Weigel, a historian of media technologies, has pointed out, Karp not only didn’t do his dissertation under Habermas, he didn’t even study in the same department.

That sort of exaggeration could be chalked up to Silicon Valley bravado. Yet it bears an eerie similarity to the bin Laden story, a rumor Palantir has allowed, or even encouraged, to be repeated as fact. As the company goes public, however, it will be required to open its books, and the facts will become inescapable. As recently as 2018, Palantir was being cited as having a valuation of $20 billion, ranking it among Silicon Valley’s top-five unicorns, alongside Uber, Airbnb, SpaceX, and WeWork. Since then, WeWork has imploded, Uber is trading below its public-offering price, and Airbnb has been hit by a pandemic-driven collapse in bookings. While Palantir’s filings indicate it hopes to surpass $20 billion, some industry analysts suspect that goal is far beyond what its business model can justify.

How much Palantir is worth depends in large part on what kind of company you think it is. Palantir markets itself as “software as a service” — a business category that includes products like Microsoft Office 365. But Microsoft is not embedding software engineers in Afghanistan to help soldiers with Excel. If, on the other hand, Palantir is more like a traditional government IT contractor, which provides people as a service, it would be valued at about one times revenue — a number that would place its current worth at less than $1 billion. “You have to make sure that you can get to a dramatically different model at Palantir … to get a valuation up to the kinds of numbers that we’re hearing out there,” Douglas Harned, a Bernstein analyst, explained in a recent webinar.

So why are people still so excited about Palantir? One former national-security official told me the company is now famous for being famous, sort of like the Kardashians. But he’s doubtful Palantir’s technology can match the sky-high valuations that came with all the hype. “As soon as there’s an IPO, I will short the stock,” he said. “If I’m right — if, in fact, Palantir is loved in the way the Kardashians are loved — well, the Kardashians are not going to be famous forever. So short the stock while they’re famous — and just wait for their 15 minutes of fame to end.”

*This article appears in the September 28, 2020, issue of New York Magazine. Subscribe Now!

Is Palantir’s Crystal Ball Just Smoke and Mirrors?