Palantir Technologies will not help you share, message, pin, post or chat. It does not exist to make you more social or connected, or even to help advertisers get to you. Its technology is deeply geeky, its work secretive. Nonetheless, it is one of the most valuable private tech companies in Silicon Valley.
Founded in 2004, in part with $2 million (€1.5 million) from the Central Intelligence Agency's venture capital arm, Palantir makes software that has illuminated terrorist networks and figured out safe driving routes through a wartorn Baghdad. It has also tracked car thieves, helped in disaster recovery and traced salmonella outbreaks. US lawyers deployed its technology against hedge fund SAC Capital, which was also an early investor in the company. (SAC, which changed its name to Point72 Asset Management after it pleaded guilty to insider trading charges, declined to comment on its investment.)
Palantir's software has been used at JPMorgan Chase to spot cyberfraud and to sell foreclosed homes; at Bridgewater Associates to help figure out investments for its $157 billion under management; and at Hershey to increase chocolate profits. The technology is complex, but the premise is simple: The software consumes huge amounts of data – from local rainfall totals to bank transactions – mashes it together and makes conclusions based on those unlikely combinations. Where is a terrorist attack likely to occur? What is a bad financial bet?
This year, Palantir, which is based in Palo Alto, California, is expected to bring in about $1 billion in revenue, mostly from private companies interested in adaptations of its intelligence software. Although it is not yet profitable, investors have given Palantir almost $900 million in total. The most recent round, last December, sold shares in the company to investors at an implied valuation for the firm of $9 billion.
All of this has its investors, including some of the world’s most successful hedge funds, salivating for a big payday from an initial public offering. “The company has been incredibly successful, and every investor likes when companies go public,” says Justin Fishner-Wolfson of 137 Ventures, which is one of those investors.
You would think this would be exciting for Alex Karp (46), Palantir's co-founder and chief executive. But Karp, who has a doctorate in philosophy and an idealistic vision for the company, says he is resisting the big wealth and publicity of an IPO. He fears the money – and the emphasis on stock price – will destroy Palantir's mission: to use its software to improve the world.
An IPO “is corrosive to our culture, corrosive to our outcomes”, he says. Karp holds about 10 per cent of Palantir. That makes him nearly a billionaire on paper, but he rents a townhouse, his only permanent residence, near the company’s headquarters. He doesn’t know how to drive, either. His backers and Palantir’s clients, he adds, “do not see us as supernormal”.
“When you are saving the world, fighting fraud and slave labour, you can do great things,” Karp says.
Palantir does not charge for most humanitarian work, which is a source of internal pride. “What concerns me,” he says, “is working with commercial entities, and non-US governments.”
There is the rub. Karp is eloquent on the subject of Palantir's saving lives, but it is in business to make money – as are its eager investors. It actively seeks corporate contracts worth tens of millions and is getting bigger by the day. People familiar with the company say it has big deals with insurers, healthcare firms and media corporations. Its advisers include Democratic strategist James Carville; former US secretary of state Condoleezza Rice; former CIA director George Tenet; and former head of Disney Michael Ovitz.
Palantir is growing by selling software to private companies. Whether those companies will properly use its privacy safeguards, which were designed for the government, is up to those customers. Palantir has worked to recover from its own ethical lapses, but Karp acknowledges it cannot control the ethics of its customers.
Still, Palantir stands out in the tech industry. Near Karp’s office are books that employees are encouraged to read on subjects such as database structures, critical histories of the CIA and improvisational theatre. The company caps salaries at $137,000 a year and bonuses at $15,000; it allows internal stock sales but only up to $300,000 a year. Given the Silicon Valley gold rush, where mundane work can make people millionaires before 30, a place where engineers settle for a “mere” total of $450,000 seems extraordinary.
Instead of selling stock to the public, Karp and other executives are toying with the idea of creating new kinds of financial instruments, like a bond that pays off on future earnings, to unlock a bit of Palantir’s value. “You may not get rich” working at Palantir, Karp says, but “you live like the prince of a small municipality”. That means not just the usual free food, drinks and foosball that are expected in Palo Alto firms but “interesting work that matters”.
Palantir is not the first company dealing with big data that has been conflicted between ideals and commerce. As graduate students at Stanford,
wrote that “advertising-funded search engines will be inherently biased toward the advertisers”. Then they started
, which makes money from advertising.
used to talk about a society of complete openness, while mining people’s lives to sell ads.
The issues at Palantir go beyond advertising. Critics are sceptical of the company’s self-aggrandising promises of privacy protection. But executives worry more about what compromise might do to the company – and to society.
Courtney Bowman, a former Google employee, works at a Palantir as a "civil liberties engineer", examining its obligations and making recommendations about how lawmakers should think about modern technology. Palantir also sponsors conferences on privacy in the age of big data. "I was a quantitative analyst at Google, doing ad auction design and targeting," he says.
“I had access to ways of deriving personal identity information without breaking any laws. It was a constant anxiety to me.”
A feature of Palantir software is a series of safeguards limiting who can see particular data, along with “audit trails” that enable investigators to check that the rules were followed. While audit-trail technology is built inside private-sector versions of Palantir, its use is not mandated. “What keeps me up at night is that we have to keep thinking about this as we grow into new markets and new regions,” Bowman says. “As you move into higher levels of computing complexity, you can’t retreat into the argument that this [the technology of finding hidden things] is neutral.”
To Palantir's critics, such introspection is a collective delusion at best. "They are a key force in the surveillance-industrial complex, but they are in denial about it," says Christopher Soghoian, a technologist at the American Civil Liberties Union.
“Their product is used by state and local law enforcement. It’s problematic for them to spread money to their critics at conferences.”
In 2011, the world got a taste of what could go wrong with Palantir's confluence of commerce and surveillance. Along with two intelligence firms, a Palantir employee had pitched a Washington law firm on ways that it could expose the workings of WikiLeaks, the group that publishes secret government and private-sector information. The pitch included the idea of using disinformation and cyberattacks.
The idea fizzled, but Anonymous, the loosely associated network of cyberactivists, posted both the pitch and emails indicating that Palantir also proposed creating misinformation about journalists, including Glenn Greenwald, who wrote in support of WikiLeaks and who recently shared a Pulitzer Prize for his articles on Edward Snowden's leaking of National Security Agency spying documents.
Karp publicly apologised to Greenwald. On the recommendation of an outside law firm, the employee was suspended for a while but still works at Palantir. Inside Palantir, people were equally troubled by published emails in which the employee compared the law-firm work to “money falling from the sky for those of us used to working for the government sector”.
If Palantir had carried out the proposed action on WikiLeaks, "we would have collapsed", says Ari Gesher, who serves as a recruiter and public face of Palantir. "We really learned that we do work in areas where we come into contact with bad actors. We have to be vigilant."
Back then, Palantir had a few hundred employees, and a culture of internal discussion. There are still such talks, but the company is now much larger – making a consensus harder to establish. Some employees do not want Palantir aiding Israel, because they disagree with its policies towards Palestinians. There are still contracts with the Israeli government. Palantir has decided not to work with China. After an internal debate, it also decided not to do business with tobacco firms.
Palantir’s products do help the US military kill people, Karp agrees, but only those with whom the nation is at war. Palantir is “building something for the betterment of the world”, he says, “but not in absence of realities about the world”. As Palantir expands into offering services to the private sector – now perhaps 70 per cent of its business – Karp’s worry is losing control of its software. The privacy controls are, after all, optional. And, ultimately, it can’t control who gets the software. If a tobacco company wanted Palantir technology, it could acquire a Palantir client.
“The thing Alex worries about the most is they have a culture that’s hard to sustain as it grows,” says Carville, the strategist. “He talks about it: ‘If we become something besides Palantir, what are we?’”
Palantir began in the mind of Silicon Valley investor and PayPal founder
. He began thinking about it in 2003, a year after he sold PayPal to
. It was two years after the terrorist attacks on the US, and Thiel wondered if the world could be made safer without losing freedoms.
“I felt we were drifting to a place in the US we’d have a lot fewer civil liberties and no real effective protection,” Thiel says.
He enlisted Karp, a college friend, as well as veterans of PayPal and his investment fund. A fan of JRR Tolkien, he named Palantir after a set of magic stones in The Lord of the Rings that grant powerful people the ability to see the truth from afar. The company headquarters are called the Shire, after the home of the Hobbits.
Palantir’s founders started with an idea from PayPal. At one point, PayPal was losing the equivalent of 150 per cent of its revenue to stolen credit-card numbers. It figured out how computers could spot activity – like a flurry of payments to a new account – at a global scale. The flagged actions would then be put before a PayPal employee to investigate. Palantir’s founders thought the same approach would work for national security. Almost no one in the venture capital world agreed except In-Q-Tel, the CIA’s venture fund, which gave Palantir $2 million.
Palantir's first full-fledged CIA job was in 2008. Karp got more work from word of mouth and donated Palantir's technology to cyberactivists, who mapped Russian hackers attacking Georgia in 2008. (The spyware was also rumoured to have found Osama bin Laden.)
To drum up private-sector business, Thiel called on Ovitz, whom he knew through Marc Andreessen, the former Netscape whiz kid turned venture capitalist. At first, Ovitz thought Palantir could be used in selling online ads, but the housing crisis changed his thinking. Banks had thousands of homes in foreclosure across the US and no idea how to efficiently clear the backlog in a collapsing market.
“The idea was to pick one bank, and the rest would follow,” Ovitz says. JPMorgan was the first. Much as Palantir figured out navigating Baghdad by analysing recent roadside attacks, satellite images and moon phases, it derived home-sale prices by looking at school enrolments, employment trends and retail sales. Data that JPMorgan thought would take two years to integrate was put into action in eight days.
JPMorgan still uses Palantir for cybersecurity, fraud detection and other work, loading half a terabyte of data on to a Palantir system each day.
Morgan Stanley is another customer. "No human can look at all the data sources at one time," says Jim Rosenthal, Morgan Stanley's chief operating officer.
The company uses Palantir to spot money laundering and employee theft, as well as for cybersecurity. Government clients also struggle with a data explosion. “It’s malpractice to have records and not search them,” says Preet Bharara, a lawyer for the southern district of New York who has used Palantir for several cases, including the SAC investigation.
Investors are growing restless. Karp says he hears from them almost every day. But while an IPO may be hard to resist forever, Thiel says Palantir “has no plans to IPO in the next few years”.
Palantir occupies about 250,000sq ft of Palo Alto buildings, which hold many of its 1,500 employees. Contracts around the world have surged as everyone's data increases in size and diversity. Karp thinks the firm can grow to 5,000 employees, without, he says, the possibly corrupting influence of going public. He hopes Palantir can save itself, along with the world. – (New York Times service)