Did Uber Steal Google’s Intellectual Property?

Silicon Valley was built on job-hopping. But when a leader of Google’s self-driving-car unit joined Uber, Google filed suit. Now the Feds are on the case.
illustration of conveyer belt of tech employees chained together
Google initiated a lawsuit after Anthony Levandowski, a former employee who helped create self-driving cars, joined Uber.Illustration by David Plunkert

In the spring of 2011, a small group of engineers working on a secretive project at Google received an e-mail from a colleague. It’s finally happening, the note read. Anthony is going to get fired. Several of the recipients gathered in one of the self-serve espresso bars that dot the company’s headquarters, and traded rumors suggesting that Anthony Levandowski—one of the company’s most talented and best-known employees—had finally gone too far.

Levandowski was a gifted engineer who frequently spoke to newspapers and magazines, including this one, about the future of robotics. On the Google campus, he was easy to pick out: he was six feet seven and wore the same drab clothes every day—jeans and a gray T-shirt—which, in Silicon Valley, signalled that he preferred to conserve his cognitive energies for loftier pursuits. Often invited to company brainstorming sessions, he was known for having a charismatic (and, to some, annoying) tendency to launch into awkward sermons about the power of technology to change the world.

It was Levandowski who, with his colleagues, had persuaded Google’s leadership to spend millions of dollars inventing self-driving cars. Google had recruited Levandowski and a handful of other roboticists four years earlier, after the group competed in the DARPA Grand Challenge, a government-sponsored self-driving race across deserts in California and Nevada. Most of the race’s competitors had built automated cars, but Levandowski had constructed a self-driving motorcycle called Ghostrider—in part, he later admitted, because he hoped that its novelty would draw attention. Although Ghostrider performed rather pitifully in its début, breaking down a few feet from the starting line, in almost every other respect it was a success: the audacity of Levandowski’s creation, coupled with his talent for charming journalists, made him the competition’s star. The National Museum of American History acquired Ghostrider for its permanent collection, and in 2007 Levandowski—then twenty-seven years old, with only a master’s degree in engineering from U.C. Berkeley—was offered a job at Google worth millions of dollars.

At the time, the company was hoping to dominate the market for navigational services with software that offered turn-by-turn instructions to urbanites seeking the quickest route to the grocery store or the gym. Google was betting that, as smartphones matured, users would willingly hand over digital information about where they were and where they wanted to go—a valuable trove for a company devoted to selling targeted ads. To perfect such software, Google needed on-the-ground details: the exact locations of speed-limit signs on roads; eye-level assessments of which off-ramps were easy to negotiate and which required sudden lane changes. Levandowski and his Grand Challenge teammates had developed a method for inexpensively stitching together thousands of landscape photographs, then combining them with G.P.S. coördinates, in order to plot navigable self-driving paths over dusty hills and creek beds. This technology could be adapted to map city streets, but millions of up-to-date photographs would have to be taken first. After Levandowski arrived at Google, his plan was to send out hundreds of cars, equipped with cameras, to photograph America’s roads. Then he encountered Google’s bureaucracy.

The company was less than a decade old, but it had almost seventeen thousand employees, including a thick layer of middle managers. Levandowski recently told me, “One of the reasons they wanted us was because Larry Page knew we were scrappy—we would cut through red tape.” Page, Google’s co-founder and chief executive, often complained that the company had become bloated, and had lost the hacker mentality that had fuelled its initial success. By the time Levandowski arrived, Google’s apparatchiks were in ascent.

“Hiring could take months,” Levandowski told me. “There was a program called WorkforceLogic, and just getting people into the system was super-complicated. And so, one day, I put ads on Craigslist looking for drivers, and basically hired anyone who seemed competent, and then paid them out of my own pocket. It became known as AnthonyforceLogic.” Around this time, Levandowski went to an auto dealership and bought more than a hundred cars. One of his managers from that period told me, “When we got his expense report, it was equal to something like all the travel expenses of every other Google employee in his division combined. The accountants were, like, ‘What the hell?’ But Larry said, ‘Pay it,’ and so we did. Larry wanted people who could ignore obstacles and could show everyone that you could do something that seemed impossible if you looked for work-arounds.”

Levandowski and his team were asked to map a million miles of U.S. roads within a year. They finished in nine months, and then set up an enormous office in Hyderabad, India, to begin mapping every street on earth. (Today, Google Maps is the dominant navigation app, used daily by more than thirty million people.) Levandowski and his boss, Sebastian Thrun—another Grand Challenge alumnus—then proposed to Google’s leadership that the next step was developing self-driving cars. In 2009, a small team of engineers, led by Thrun, was assigned to a secret self-driving-car division, which was given the code name Project Chauffeur. Levandowski’s focus would be hardware development.

One of the group’s first objectives was figuring out how to give an autonomous car “eyes”: technology connecting lasers and cameras mounted on the roof to an onboard computer that visualized the road ahead, from traffic signs to pedestrians and other automobiles. Such systems had been created before, at universities, and each had taken years to build. Levandowski again devised a work-around. After joining Google, he had, on the side, created two independent companies, 510 Systems and Anthony’s Robots, which held the rights to technologies that he had developed for his autonomous motorcycle and other outside projects. As he saw it, Google could potentially skip years of redundant research if Project Chauffeur simply bought the necessary hardware from his firms. In effect, Levandowski was proposing to acquire crucial technology from himself, and pay for it with Google’s money.

Google said yes. Although some executives were aware of this unusual arrangement, others were in the dark. “At first, no one really understood that Anthony was selling us his own stuff, but people eventually figured it out,” one of Levandowski’s former colleagues told me. “It seemed shady, but at the same time everyone wanted to move as fast as possible, and this was an easy solution, so we didn’t ask a lot of questions. That ended up being a mistake.”

During the next few years, Project Chauffeur grew to hundreds of employees, and Google spent a small fortune building a fleet of self-driving cars. Transportation is one of the world’s largest industries. If Google could become the first company to bring autonomous-vehicle technology to the marketplace, the breakthrough would be worth billions. Levandowski was central to Google’s plans, but, as Project Chauffeur expanded, his leadership style became increasingly divisive. He was adept at solving problems and at rallying workers, but he was brusque and obsessive, prone to belittle teammates who disagreed with him.

“The five-second rule doesn’t apply to soup.”

He also seemed preoccupied by his personal compensation. “We were once driving to a meeting together, and we were talking about how much we wanted to earn from Chauffeur,” one of Levandowski’s co-workers said. “I told him I wanted to make a hundred million dollars, which seemed like a totally inconceivable figure to me. And—I remember this very clearly—Anthony looked over, with this pitying expression, and said I was thinking way too small. He said he expected to make a billion dollars, at least. This technology was going to change the world, and a billion was the minimum of what he deserved.” Levandowski sometimes wore a custom-designed gray T-shirt, a gift from a colleague, that read “I Drink Your Milkshake”—a line from “There Will Be Blood,” Paul Thomas Anderson’s film about a murderously ambitious oilman. “He was that kind of guy,” the co-worker said. “You know, an asshole. But a really gifted one. Our asshole, I guess.”

In 2010, employees began to hear rumors that Levandowski, on behalf of his outside companies, had met with some of Google’s competitors, including a firm that was working with Microsoft, to sell them the same navigational technology that Google was paying to use. Levandowski later insisted that he had never hidden the fact that his companies had relationships with other firms, and argued that Google had never acquired an exclusive license. To Levandowski’s teammates, however, it felt like a betrayal. His bosses launched an internal inquiry. When Levandowski was questioned, he said that it was not his intention to help other companies compete with Google, but noted that he was contemplating leaving in order to focus on 510 Systems and Anthony’s Robots, which might become competitors of their own in the self-driving-car business. That’s when the whispers about firing him began.

However, according to former Google executives, when Page heard that Levandowski might be fired, or that he might leave the company, he ordered a reversal. Google needed people like him. Many of its major innovations, like Gmail and AdSense, were more than half a decade old, an eternity in the tech industry. Various “moonshots”—attempts to diversify Google’s revenue stream by building new businesses—had consumed billions of dollars but shown few significant results. The company had spent hundreds of millions of dollars developing Google Glass—augmented-reality spectacles unveiled, to great fanfare, in 2012—but had quietly shuffled them aside when their performance proved underwhelming and glitchy. Even when Google acquired innovative startups—such as the smart-thermostat manufacturer Nest, in 2014—the pace of invention often stalled almost as soon as the purchase closed.

Page ordered his lieutenants to negotiate the purchase of 510 Systems and Anthony’s Robots, and to consider giving Levandowski an expanded leadership role. “Anthony became who he is because Larry nurtured and protected him,” a former high-ranking Google executive told me. “They were friends—they liked having dinner and geeking out.”

Within Project Chauffeur, though, there was considerable resistance to elevating Levandowski. Thrun, in an e-mail to his colleagues, said that several team members had “concerns about Anthony’s commitment and integrity.” Another executive, David Lawee, wrote that, even if Google was ready to “take the risk with Anthony” and make the acquisition, “I can say, definitively, that if I was choosing a business partner to start a company with, there is no way in hell that I would proceed.”

Page was adamant. According to internal Google e-mails, he ordered executives to “make Anthony rich if Chauffeur succeeds.” Two months later, Google bought 510 Systems for twenty-two million dollars. It also purchased Anthony’s Robots; in return, Levandowski was guaranteed a future payment tied to the total value of Project Chauffeur. Google agreed to give him a claim on ten per cent of the division’s eventual worth—a kind of shadow equity that would vest in four years. The stake eventually paid him more than a hundred and twenty million dollars, one of the largest such payouts in Google’s history.

“Sometimes the best way to get promoted at Google is to threaten that you’ll quit,” Levandowski told me. “If you’re, like, ‘Hey, I’m going to do my own startup,’ they’re, like, ‘O.K., we need to buy you.’ That’s how Silicon Valley works. You have to shake things up, create pressure. The people who win here are the ones who believe in the biggest future and are willing to take risks to make it come true.”

Silicon Valley has always been built as much on treachery as on innovation. The interplay can be traced back to 1957, when the Valley was essentially created by a group of young engineers who gathered to discuss betraying their boss. The engineers worked for William Shockley—a recipient of the Nobel Prize who’d helped invent the transistor, and a disastrous supervisor. He had founded a small company amid the fruit farms of Mountain View, California, choosing the location in part for its proximity to his mother. Its offices were less than three miles from where Google is based today.

Shockley had recruited a small team—many of them recent graduates from Stanford and M.I.T.—to manufacture some of the first semiconductors. Within a year, most of his employees were ready to quit. Shockley was a paranoid micromanager, and a racist prone to temper tantrums; he later donated his sperm to science, with the stipulation that it be used to inseminate only women who qualified for Mensa. His suspicions of his employees were so extreme that he once proposed that they all take lie-detector tests. When eight of Shockley’s top engineers discovered that they had similar misgivings about their leader, they approached a corporation called Fairchild, which offered to help them create a new company. After the eight informed Shockley that they were leaving, his lieutenant told them that it felt as if “a good friend has stabbed him in the back.”

The defectors came to be known as the Traitorous Eight, and though they hated the name, it became a rallying cry for a subsequent string of entrepreneurial betrayals that built Silicon Valley. Fairchild Semiconductor, as the Traitorous Eight’s new company was called, made its founders rich, introduced the phrase “silicon wafer” to the world, and trained a generation of computer engineers in the fine art of disloyalty. Before long, Fairchild employees were departing to found new firms; more than a hundred were created, including Teledyne and Intel.

At the time of the Shockley betrayals, other regions were much better positioned than Silicon Valley to become the tech industry’s breadbasket. Route 128, a beltway around Boston, was home to so many universities and pioneering computer firms that it soon became known as America’s Technology Highway. But by the nineteen-seventies Route 128 had been overshadowed by Northern California. Economists later suggested that the Valley’s culture of betrayal was a major reason for its success. Massachusetts’ laws made it difficult for employees to join rival companies or create new businesses. Engineers in Boston were typically forced to sign non-compete agreements that required them, if they quit, to wait at least a year before joining a competitor or creating their own firms. But in California non-compete agreements were illegal. That prohibition had been inserted into the state’s commercial code almost by accident, in the eighteen-seventies, when California lawmakers—seeking to save time—virtually copied a set of statutes that had been proposed (and then rejected) by New York’s legislature. When California’s early legislators outlawed “every contract by which anyone is restrained from engaging in a lawful profession,” none of them could have foreseen that, a century later, their decision would transform the global economy.

As companies in Silicon Valley multiplied, you could “quit your job on Friday and have another job on Monday,” a California engineer told the economist AnnaLee Saxenian. “You didn’t necessarily even have to tell your wife. You just drove off in another direction on Monday morning.” New businesses emerged across the state at a dizzying rate. Every year in the nineteen-seventies, a third of California’s tech workers quit their jobs and joined competitors, or launched new companies.

“The career paths of Silicon Valley engineers and managers resembled Brownian motion,” Ronald J. Gilson wrote, in a 1999 law-review article. “They moved between companies, founded startups, supplied former employers, purchased from former employees, and in the course of their careers developed personal and professional relationships that cut across companies and competition.” As these motes danced across California’s business landscape, they spread secrets and shared insider knowledge. Such dissemination, known among economists as “knowledge spillover,” spurs innovation. “Why did Adidas build a workshop in Portland?” Sharon Sandeen, an intellectual-property expert at Mitchell Hamline School of Law, said to me recently. “They could go anywhere in the world.” But Nike is based in the Portland area, and Adidas, by establishing an office there, could more easily hire former Nike employees, thereby acquiring knowledge of its rival’s methods. “Ideas are going to leak,” Sandeen said.

Recent studies suggest that, in some industries, allowing employees to move freely among companies can dramatically increase the pace of innovation. Knowledge spillover helps companies avoid repeating the mistakes of their competitors. It educates a dispersed cohort of researchers about new discoveries. Knowing that your rival will soon hire your best employees, and learn all your secrets, encourages firms to avoid resting on their laurels. As a former high-ranking Google executive told me, “We want people to try crazy things and then worry that someone else is going to take that idea unless they run as hard as they can to use it first.”

In recent years, however, innovation within Silicon Valley appears to have slowed. Scholars note that many of the largest companies—including Google, Apple, Facebook, and Microsoft—haven’t released revolutionary products in more than a decade. The iPhone débuted in 2007. Facebook’s News Feed was introduced in 2006. Microsoft’s Xbox is sixteen years old. And, when disruptive competitors—such as Instagram, WhatsApp, and Waze—have emerged, larger firms have usually acquired them. Big Silicon Valley companies now assume that many employees will stick around for decades.

In the increasingly sclerotic world of Silicon Valley, Levandowski was an outlier. The former high-ranking Google executive said, “Most people basically stop giving a fuck once they’re rich. And that’s why Anthony was so special. He never relaxed. He was one of those guys who would work and work and work, and then pass out, get up, and work some more, just pushing, constantly, until he made an impossible idea into a reality.”

Even after Levandowski was promised more than a hundred million dollars, he continued urging himself to “think outside the box”—that Silicon Valley cliché for transcending cliché. “And that was good and bad,” the executive said. “Because there are times when not seeing the box helps you do the impossible, and times when acknowledging that the box exists is what keeps us from doing something dumb—or going to jail.”

After Google acquired Levandowski’s companies, Project Chauffeur continued to move forward, and Google’s investment in autonomous technologies soon surpassed a billion dollars. Residents of Northern California began seeing self-driving cars cruising through their neighborhoods—Priuses and Lexuses with spinning gizmos on their roofs and a tendency to stop hesitantly at intersections, as if student drivers were behind the wheel. In 2015, Project Chauffeur celebrated more than a million self-driven miles.

“Hold on—I’m getting information as to why these two should not be wed.”

That year, Levandowski began inviting select colleagues to talk with him outside the office. Just as he had once started Anthony’s Robots on the side, he was thinking of founding another independent firm. It had been almost four years since he sold his companies to Google, and his payouts were about to vest, giving him a fortune that would be vast even by Silicon Valley standards. But Levandowski, who now had two children, told colleagues that he wasn’t about to become complacent. He was unhappy at Google, and felt that the pace of progress at Project Chauffeur had stalled, in part because of arguments among executives.

Some of the biggest fights involved risks that Levandowski was taking in self-driving experiments. The software that guided Google’s autonomous vehicles improved by ingesting immense amounts of test-drive data. One effective way to teach autonomous vehicles how to, say, merge onto a busy freeway is to have them do so repeatedly, allowing their algorithms to explore various approaches and learn from mistakes. A human “safety driver” always sat in the front seat of an autonomous vehicle, ready to take over if an experiment went awry. But pushing the technology’s boundaries required exposing the cars’ software to tricky situations. “If it is your job to advance technology, safety cannot be your No. 1 concern,” Levandowski told me. “If it is, you’ll never do anything. It’s always safer to leave the car in the driveway. You’ll never learn from a real mistake.”

One day in 2011, a Google executive named Isaac Taylor learned that, while he was on paternity leave, Levandowski had modified the cars’ software so that he could take them on otherwise forbidden routes. A Google executive recalls witnessing Taylor and Levandowski shouting at each other. Levandowski told Taylor that the only way to show him why his approach was necessary was to take a ride together. The men, both still furious, jumped into a self-driving Prius and headed off.

The car went onto a freeway, where it travelled past an on-ramp. According to people with knowledge of events that day, the Prius accidentally boxed in another vehicle, a Camry. A human driver could easily have handled the situation by slowing down and letting the Camry merge into traffic, but Google’s software wasn’t prepared for this scenario. The cars continued speeding down the freeway side by side. The Camry’s driver jerked his car onto the right shoulder. Then, apparently trying to avoid a guardrail, he veered to the left; the Camry pinwheeled across the freeway and into the median. Levandowski, who was acting as the safety driver, swerved hard to avoid colliding with the Camry, causing Taylor to injure his spine so severely that he eventually required multiple surgeries.

The Prius regained control and turned a corner on the freeway, leaving the Camry behind. Levandowski and Taylor didn’t know how badly damaged the Camry was. They didn’t go back to check on the other driver or to see if anyone else had been hurt. Neither they nor other Google executives made inquiries with the authorities. The police were not informed that a self-driving algorithm had contributed to the accident.

Levandowski, rather than being cowed by the incident, later defended it as an invaluable source of data, an opportunity to learn how to avoid similar mistakes. He sent colleagues an e-mail with video of the near-collision. Its subject line was “Prius vs. Camry.” (Google refused to show me a copy of the video or to divulge the exact date and location of the incident.) He remained in his leadership role and continued taking cars on non-official routes.

According to former Google executives, in Project Chauffeur’s early years there were more than a dozen accidents, at least three of which were serious. One of Google’s first test cars, nicknamed KITT, was rear-ended by a pickup truck after it braked suddenly, because it couldn’t distinguish between a yellow and a red traffic light. Two of the Google employees who were in the car later sought medical treatment. A former Google executive told me that the driver of the pickup, whose family was in the truck, was unlicensed, and asked the company not to contact insurers. KITT’s rear was crushed badly enough that it was permanently taken off the road.

In response to questions about these incidents, Google’s self-driving unit disputed that its cars are unsafe. “Safety is our highest priority as we test and develop our technology,” a spokesperson wrote to me. The company said that, in the case of the KITT collision, a report was submitted to the authorities, and that although multiple participants later sought medical care, “every person involved left the scene on their own accord.” As for the Camry incident, the spokesperson described it as “an unfortunate single-car accident in which another car failed to yield to traffic”; because Google’s self-driving car did not directly hit the Camry, Google did not cause the accident.

Since 2014, California regulations have required companies to report any instance in which a self-driving vehicle is “in any manner involved in a collision originating from the operation of the autonomous vehicle on a public road that resulted in the damage of property or in bodily injury or death.” The Camry accident occurred three years before this regulation was passed; since the rule went into effect, Google has reported thirty-six additional accidents. If Google is still failing to report accidents in which its cars did not hit other vehicles, then there may have been more undocumented incidents. “There’s lots of times something happened because one of our cars drove erratically but never hit anyone,” a former senior Google executive told me. Google cars sometimes stopped suddenly, including at intersections, causing other cars to swerve. (A spokesperson for Google declined to discuss the company’s reporting policies.)

Phil Ting, a Democrat in the California state legislature who represents San Francisco, has sponsored autonomous-vehicle legislation. “If there’s a self-driving car behaving in an unsafe way, that obviously should be reported,” he said. “If they aren’t being reported, that deserves examination.” The former senior Google executive said, of accidents involving the public, “That’s the Silicon Valley way, to fail fast and fail often. But these are cars we’re talking about, not iPhone apps. The wrong failure can kill someone.”

In 2015, Levandowski began hosting events for colleagues, including a paella dinner, at which he described his latest vision: he wanted to create a new self-driving company that would focus on autonomous trucking. He said that this business model wouldn’t compete with Google. And, even if it did, he lived in California, and there was nothing in state law to stop him. He had already spoken to potential investors, including Uber, which had launched its own self-driving unit.

When Google executives learned that he was recruiting employees to leave, they were livid. “We need to fire Anthony Levandowski,” Chris Urmson, the head of Project Chauffeur, e-mailed his colleagues on August 4, 2015. “I have just heard today from two different sources that Anthony is approaching members of their team attempting to set up a package deal of people that he could sell en masse to Uber.” Once again, however, Google’s top executives, including Page, stepped in to protect Levandowski. Page’s decision was perhaps inflected with self-interest: he had recently founded an outside project of his own—a flying-car enterprise called Kitty Hawk—and had asked Levandowski to help him in his spare time. Other Google executives worried that, if Levandowski was fired, some of his allies in the hardware division might leave with him. Urmson was directed to do what was necessary to keep Levandowski.

Urmson tried to reach a détente with him, but it was too late. In January, 2016, Levandowski e-mailed Page to announce that he was resigning, noting, “I want to be in the driver seat, not the passenger seat, and right now [it] feels like I’m in the trunk.” Urmson tracked down Levandowski and marched him off the property. Many of Levandowski’s teammates already had job offers from his new startup, and Urmson spent the rest of the week offering them tens of thousands of dollars if they agreed to remain at Google. Some accepted; others declined. Half a dozen Google employees eventually left to join Levandowski.

Rumors soon began circulating that Uber was going to buy Levandowski’s company, which he had named Ottomotto, and by the end of February, 2016, a term sheet was in place. Uber said that it was prepared to hand over a one-per-cent stake in the company—at the time, the equivalent of six hundred million dollars—in exchange for Ottomotto. “That really got everyone’s attention,” a former Google engineer told me. “If Anthony could get that much, how much could we get?” In August, Project Chauffeur’s head of software wrote to colleagues, “There are a bunch of people who were considering joining Otto earlier this year.” The hundreds of millions of dollars being handed out by Uber was “having a strong impression,” he noted, adding, “We have another wave of good people reconsidering their options.”

Project Chauffeur employees began calling in sick so that they could interview with other firms or with venture capitalists. People started paying close attention to their co-workers’ shoes. “The only reason an engineer buys new, expensive sneakers is if they’re looking for a new job,” one person told me. Project Chauffeur’s hallways were suddenly filled with unscuffed Pumas and Allbirds. A stream of employees joined competing robotics startups. Even Urmson left and launched his own firm.

Google’s leaders were particularly concerned about defections to Uber. “Everyone is terrified of Uber,” the former high-ranking Google executive said. “Uber is the first major Silicon Valley firm with your permission to know who you are, where you’re going, who you’re with, what businesses you visit regularly, and they have your credit-card information on file. And then, when they hired Anthony, they suddenly got access to someone who could help them skip over a billion dollars’ worth of mistakes.”

By the summer of 2016, Google was looking for a way to stop Levandowski’s deal. That August, the agenda for a meeting of executives included a hastily written game plan: “Look at prior to anthony’s departure, in course of his employment, whether or not they took with them confidential information or trade secrets of the business, violated their non-solicitation agreement, to what extent information misappropriated.”

Soon afterward, forensic engineers, led by the Google executive Gary Brown, began combing Project Chauffeur’s databases in search of material that could be used to block Uber’s acquisition of Ottomotto. They eventually discovered a small lead. According to Google, a month before Levandowski resigned, he had plugged his work-issued laptop into a Google server and downloaded about fourteen thousand files, including hardware schematics. He transferred the files to an external drive and then wiped his laptop clean. Lawyers later learned that, around the same time, an engineer who had left with Levandowski, Lior Ron, had conducted Internet searches for “how to secretly delete files mac” and “how to permanently delete google drive files from my computer.” (Ron declined to comment.) Lawyers later saw a chat message that Levandowski sent to Ron several weeks after he left Google: “Make sure you delete all the messages tonight on both your PC and iPhone.” This was evidence, Google felt, that Levandowski had exploited Project Chauffeur’s secrets to jump-start Ottomotto.

But, when lawyers asked Google engineers to evaluate what Levandowski had taken, one of them dismissed it as “low value” information; he told a Google attorney, in an e-mail, that “it makes me uncomfortable to think that lawyers are trying to ascribe suspicion to” the downloads.

In December, 2016, Google spun its self-driving unit into a stand-alone division called Waymo, which, employees were told, stood for “a new way forward in mobility.” That month, Waymo’s lawyers learned that a mid-level worker had accidentally been forwarded an e-mail from an outside vender, a company called Gorilla Circuits, which had been hired by Uber to manufacture circuit boards for self-driving cars. As Waymo lawyers later wrote, the drawings “bore a striking resemblance to—and shared several unique characteristics with—Waymo’s highly confidential . . . circuit boards, the design of which had been among the more than 14,000 files downloaded by Mr. Levandowski.”

On February 23, 2017, Waymo filed suit in a federal court, alleging that Levandowski had absconded with information of nearly incalculable worth. The suit demanded $1.85 billion in damages, and sought to prevent Levandowski from using Waymo’s proprietary information at Uber, Ottomotto, or any other self-driving company. Waymo could not claim that Levandowski had been bound by a non-compete clause—and it was prohibited from suing him directly, because his employment contract had contained an arbitration provision. So it sued Uber and Ottomotto instead, under recently passed federal legislation that offered a novel angle of attack: Levandowski, Waymo claimed, had committed trade-secret theft.

A bargain lies at the heart of America’s intellectual-property laws: if you are willing to publicly describe a new invention with enough precision that others can copy it, then the government will give you a patent, a temporary monopoly on the invention’s use. “Patent and copyright laws are intentionally made ‘leaky,’ ” as the Yale Journal on Regulation eloquently put it. The government allows you to profit from your hard work while still insuring that others can learn from your discoveries.

But what if the company makes a discovery and does not want to share it? In that case, if a company goes to reasonable lengths to keep its intellectual property confidential, a court might recognize it as a “trade secret”—and prohibit employees from taking those ideas to new workplaces. Each state has its own trade-secret laws, but relying on them in lawsuits can be risky, in part because in most cases a judge or a jury—people who typically have little technical training—determines whether a bit of information qualifies as a trade secret, and if a court rules against you there’s nothing to stop others from taking your idea.

Almost anything can be claimed as a trade secret: a manufacturing technique, a recipe, a bureaucratic process. The fact that trade secrets are hard to define breeds paranoia. Do we own the knowledge inside our heads, or can previous employers lay claim to our memories? If you remember that certain experiments yielded no useful results, must you repeat them at a new workplace, lest your old bosses claim theft of “negative know-how”? If trade-secret laws are enforced too strictly, the practice can undermine employee mobility, destroying the Brownian motion that fosters knowledge spillover.

Large tech companies have often battled over patents, but have typically shied away from trade-secret disputes, which, among other things, can so easily destroy reputations. “With trade-secret theft, there’s this moral component,” Michael A. Jacobs, a technology attorney who has worked with Uber and other Silicon Valley companies, said. A few decades ago, Jacobs sued a tech company that was accused of patent infringement, and he wasn’t entirely confident that he would win the case. Then, one night, Jacobs held his client’s microchip schematics up to the light and overlaid it with a schematic from the defendant. The lines matched exactly. The defendant had once worked for Jacobs’s client, and had apparently photocopied some of his boss’s drawings and used them as the basis for his own designs. Jacobs said, “Because it was a photocopy, we had them over the barrel. It opened the possibility of a trade-secret claim.” The defendant, terrified of “becoming a pariah,” settled immediately.

In recent years, trade-secret lawsuits have become more common, largely because of changes in federal law. In 1996, Congress, in an attempt to combat foreign theft of American intellectual property, passed the Economic Espionage Act, which made it easier to prosecute people who steal corporate secrets. (Concerned about Chinese thefts, the F.B.I. created promotional materials warning American businessmen travelling to Asia to beware of amorous “honey pots” who had sophisticated hacking techniques.) In 2016, Congress passed the Defend Trade Secrets Act, which allowed companies to sue in federal civil court over trade-secret misappropriation, and to block employees accused of taking trade secrets from working at competing firms. This little-tested law is what Google used to target Levandowski.

The federal legislation represents “a huge expansion of trade-secret law,” Orly Lobel, a professor at the University of San Diego, told me. She added that it “advantages large, incumbent companies, because they’re the ones with the money and the resources to sue, and now they can imply that, if they win against you, you might go to jail.”

One reason to worry that big Silicon Valley firms will exploit these new federal laws is the fact that they have already resorted to underhanded tactics to prevent employees from finding new jobs. In the mid-two-thousands, some of the largest tech firms struck secret “no poaching” agreements. Google and Apple were particularly egregious. “If you hire a single one of these people, that means war,” Steve Jobs told a Google executive in 2005, after a recruiter approached Apple employees. Jobs soon received an e-mail saying that Google’s chief executive had promised he’d “firmly stopped all efforts to recruit anyone from Apple.” Government antitrust lawyers sued Google, Apple, and several other tech companies, arguing that their collusion had “substantially diminished competition” and illegally deprived workers of job opportunities. The companies settled with the government, and, by 2015, had paid more than four hundred million dollars to settle related civil claims.

Now that the 2016 law has been passed, however, tech giants no longer need secret pacts. They can simply scare employees into thinking that, if they leave, they might find themselves in federal court, accused of trade-secret misappropriation. Since the law was enacted, the number of federal trade-secret suits has skyrocketed; more than eleven hundred were filed last year, most of them by large companies against employees who went to work for other American firms. There have been more cases in California than in any other state.

One of the biggest impacts of the new laws is a cultural shift among federal prosecutors, who, rather than waiting for clear evidence of criminal wrongdoing, have begun encouraging companies to initiate, or coöperate with, criminal trade-secret investigations as soon as suspicions arise. Federal prosecutors in the Computer Hacking and Intellectual Property Unit, which is based in California, now regularly speak at technology conferences and urge attendees to hand over evidence of trade-secret misappropriation as soon as it is discovered, so that the authorities can start building indictments.

Historically, Silicon Valley companies have been wary of invitations from federal prosecutors. Companies like Apple and Google have, typically, forced the government to get search warrants, court orders, or subpoenas before handing over internal data. In 2013, Google’s chief lawyer, David Drummond, wrote on the company’s blog, “We require that government agencies conducting criminal investigations use a search warrant to compel us to provide a user’s search query information and private content stored in a Google Account.” Google had even fought a government subpoena seeking data on child-pornography searches.

But Silicon Valley executives who have become suspicious of former employees have been much more willing to work with federal officials. Federal lawyers told me that tech companies now regularly approach law enforcement in the hope of triggering criminal investigations—which can check employee mobility. “A big trade-secret case sends a message that you need to be careful about leaving, that you’re at risk, even if there’s no non-compete, because they can get you for what’s in your head,” Lobel said. She added, “I think that’s why Google went after Levandowski. They wanted to send a message. They want defections to feel like more of a risk.”

Waymo L.L.C. vs. Uber Technologies, Inc., went to trial on February 5, 2018, and by the time a jury had been selected a hundred and twenty-nine attorneys had filed more than a hundred thousand pages of pleadings, motions, affidavits, and other paperwork—a mountain of documents so overwhelming that, at one point, the judge cautioned that falling boxes might cause an injury. Attorneys from both sides told me that tens of millions of dollars were spent preparing the suit.

The months leading up to opening arguments seemed more like a soap opera than like an orderly execution of justice. Levandowski refused to discuss the case with reporters but attracted headlines with a curveball revelation: he had founded a church, called the Way of the Future, that was devoted to “the realization, acceptance, and worship of a Godhead based on Artificial Intelligence.” Machines would eventually become more powerful than humans, he proclaimed, and the members of his church would prepare themselves, intellectually and spiritually, for that momentous transition. Some people wondered if the church, as a nonprofit organization, was a scheme to protect Levandowski’s fortune, but he assured reporters that he was sincere. “I don’t believe in God,” he told me. “But I do believe that we are creating something that basically, as far as we’re concerned, will be like God to us.”

Meanwhile, in the courtroom, there were battles over who would be deposed, how long those depositions would last, and when bathroom breaks should occur. The trial itself was repeatedly delayed—once because a former Uber employee alleged, in a letter, that Uber had a clandestine unit devoted to spying on competitors, including Waymo. (After Uber agreed to pay the former employee four and a half million dollars, ostensibly for consulting services, he disavowed his claim.) Both sides leaked embarrassing details to the press. Even the Levandowskis’ nanny assumed a minor role in the drama, when she filed a multimillion-dollar suit of her own, claiming that Anthony had been emotionally abusive to her, and that he had a drawer that was filled with sex toys. (The lawsuit, which offered no evidence of her claims and contained multiple factual errors, was eventually dropped.)

The judge, William Alsup, quickly tired of such distractions. “Despite the excellent quality of the lawyers here, I cannot trust what they say,” he announced in court. The documents he was being shown, he said, included “a lot of half-truth” and arguments that were “not quite accurate.” Alsup clearly thought that something unseemly had occurred, writing in one ruling that Levandowski had resigned from Waymo “under highly suspicious circumstances,” and that the “14,000-plus purloined files likely contain at least some trade secrets.” He also noted that “it would strain credulity to imagine that Levandowski plundered Waymo’s vault the way he did with no intent to make use of the downloaded trove.” Yet Alsup wasn’t sure if Waymo had demonstrated that any of its information had been used in an illegal manner. “If you can’t prove that Uber got these trade secrets, then maybe you’re in a world of trouble,” he told Waymo’s lawyers.

“I see a series of cataclysmic events that will forever alter the course of human history, and, as usual, it will be all about you.”

The fact that Levandowski had downloaded fourteen thousand documents turned out to be less damning than had initially been assumed. The server he had plugged his computer into had been programmed to transfer files automatically, so it wasn’t unusual to see such an extensive download. Moreover, the document “trove” was considered of such little importance that Google had contemplated storing it outside the company’s servers. Before the lawsuit, one of the Google engineers asked to evaluate the downloads had said, “Doesn’t ring the alarm bells for me.” (Later, as a witness for Waymo, the engineer said that, given the context of Levandowski’s imminent departure, “maybe that’s suspicious.”)

Waymo had at first claimed that Levandowski transferred a hundred and twenty-one trade secrets to Uber, and that Uber was violating multiple patents. The Judge, however, found some of these allegations “meritless,” and concluded that Waymo had “overreached in attempting to claim ownership over general principles” of engineering. “In short, Waymo’s supposed trade secret is nothing more than Optics 101,” he declared. As the trial date drew closer, Waymo’s claims grew considerably smaller, in part because the company hoped to expedite the proceedings. By the time the jury was chosen, Levandowski and Uber stood accused of misappropriating just eight trade secrets, and all patent claims had been discarded.

The one thing that almost everyone in the courtroom seemed to agree on was that Levandowski was a troubling character. Even his new employer had renounced him. “Uber regrets ever bringing Anthony Levandowski on board,” a lawyer for the company told the jury. “All Uber has to show for Anthony Levandowski is this lawsuit.” Soon after the suit was filed, Uber fired him. The only person who had a kind word for Levandowski, it seemed, was Levandowski himself—but he wasn’t saying anything in the courtroom. He had asserted his Fifth Amendment right against self-incrimination, and had refused to answer virtually every question that lawyers asked him. He would not say what material he had taken, or if any of it had been shared. He would not confirm his ownership of outside companies or describe his conversations with any executives. At one point, Alsup made a special referral to the U.S. Attorney’s office, suggesting that federal authorities might want to explore whether Levandowski had committed any crimes.

As the trial unfolded, the wooden pews of the courtroom—crowded with journalists and lawyers—were regularly cleared out so that witnesses could offer private testimony. After all, if a trade secret is revealed in open court, it isn’t credibly a secret anymore. Over time, however, even casual observers gleaned some details. One of the alleged trade secrets concerned how diodes were positioned on a circuit board. Another had to do with the placement of small holes used to screw those circuit boards in place. Another focussed on the placement of a lens. Some of the trade secrets seemed like genuine advances that helped a self-driving car visualize its surroundings; others appeared less significant. On the second day of the trial, Alsup told Waymo’s legal team, “What you want to hide from the public does not deserve to be hidden.”

The jurors, among them a property manager who spoke limited English and a telephone-line repairman with a high-school diploma, spent much of their time looking bored or bewildered. Occasionally, they fell asleep. They were asked to examine the complex electrical schematics of a LIDAR surveying device—which uses pulses of light to ascertain how far away surrounding objects are—and determine whether a specific technique for spacing the device’s circuit boards constituted a trade secret or should be considered common knowledge among engineers.

“I’m not sure I totally understood what was going on,” one of the jurors told me after a day in court. “I wanted to get a murder trial, but I got this.”

With each passing day, Waymo’s core argument—that Levandowski had stolen valuable secrets and given them to Uber—seemed to grow weaker. Its case depended, in part, on proving that Travis Kalanick, the co-founder of Uber and its chief executive at the time Levandowski was hired, had conspired with him. Kalanick was a tempting target because he had made so many missteps—screaming at drivers, encouraging workers to flout safety regulations, ignoring employee reports of sexual harassment—that Uber had recently fired him. Waymo’s lawyer, Charles Verhoeven, had pledged in his opening statement to show that Kalanick had told his lieutenants, “We want to find cheat codes.” Verhoeven went on, “I didn’t know what a cheat code was, because I’m too old. But I understand that, when you play video games, a cheat code is something that allows you to skip what you need to do to get to the next level. Well, that’s the C.E.O. of the company saying, ‘We want to use Levandowski to find these cheat codes.’ In their own words, in written documents, Mr. Kalanick said he wanted to use Levandowski to leapfrog Google.”

But when Kalanick took the stand he came across less like a villain than like a translator helping an out-of-touch attorney understand how kids talk these days. “Cheat codes,” Kalanick explained, was a term used by people in the tech industry to describe “elegant solutions to problems that haven’t already been thought of.”

“Mr. Kalanick, I think I read in the press that you’ve been playing a lot of video games lately—is that true?” Waymo’s lawyer asked.

“iPhone games is my thing,” Kalanick said.

“In the context of video games, you know what a cheat code is, don’t you, sir?” the lawyer asked.

“Yes, but those codes in those games are put there on purpose by the publisher of those games,” Kalanick replied. “It’s just part of the fun.”

“A cheat code allows you to skip ahead and not have to do the game and do the work to get from one level to the other, yes?” the lawyer asked.

“No,” Kalanick replied, and then fell silent. The jury looked bewildered at why the courtroom had devolved into a Reddit forum. One juror, clearly a gamer, whispered to his neighbor and, as if holding a controller, mimed the start of a classic Nintendo cheat code: “up,” “up,” “down,” “down.”

“No further questions,” Waymo’s lawyer declared. Kalanick left the witness stand, and received an awkward hug from his elderly father, who had been watching. “I’m proud of you,” his dad whispered hoarsely as they walked out of the courtroom.

“That’s when I knew we were gonna be just fine,” an Uber lawyer later told me.

Before the start of the trial, Waymo had demanded $1.85 billion in damages. Now, faced with the prospect of potential defeat, executives called a meeting. One of Waymo’s lawyers told me, “If we lost the case, there was a possibility that the court could unseal our records, and the things we said were trade secrets would fall into the public domain.” Waymo executives instructed their attorneys to settle.

The next morning, Waymo’s lawyer told the Judge that the company was dropping its lawsuit, in exchange for 0.33 per cent of Uber’s equity—stock worth about two hundred and fifty million dollars. No one would have to admit serious wrongdoing. Uber promised not to use Waymo hardware or software going forward—which, given that it had said it wasn’t using any to begin with, was a relatively easy pledge, just as handing over a fraction of equity was a small sacrifice for a company valued at more than seventy billion dollars.

“If you go deep enough into the jungle, there’s always a bar scene.”

But Waymo’s sudden retreat from the civil case didn’t mean that executives had exhausted their legal options. They had also initiated a private arbitration against Levandowski, which is ongoing, seeking to reclaim his hundred-and-twenty-million-dollar bonus. And yet another avenue of attack had opened. An assistant U.S. Attorney named Matt Parrella had approached Google and Waymo and said that, if the companies would help him, he was interested in opening a criminal investigation into Levandowski.

Soon, Parrella and his colleagues had access to confidential documents that Waymo had previously sought to seal from the authorities and from the public. A lawyer with firsthand knowledge of the U.S. Attorney’s investigation said that it looked as if “Google initially didn’t want to help with a criminal investigation, but when the trial didn’t go so well they started opening up.” The lawyer continued, “It’s essentially impossible to build a trade-secret criminal case unless the victim fully coöperates. The Feds don’t know where to look unless they get a road map.” (The U.S. Attorney’s office declined to comment on the case.)

Instead of requiring “that government agencies conducting criminal investigations use a search warrant,” as Google’s Web site had proclaimed, the company allowed federal investigators to review the private testimonies and records of Waymo employees and to examine internal computer logs. In some cases, Waymo’s lawyers suggested to the U.S. Attorney’s office which documents to subpoena; this allowed the company to hand over private information to the government without officially coöperating. “It’s incredibly disappointing that Google, which was founded on the ideal of openness, has become this creepy, selfish conglomerate that coöperates with the government,” the former high-ranking Google executive told me.

A federal official who worked on trade-secret investigations said that Silicon Valley companies “couldn’t be more two-faced about” coöperating with prosecutors. “They want good P.R. But privately they’re, like, Hey, we’ll take the subpoena, and we’ll even help you write it.” And, unlike in a civil trial, federal prosecutors do not need evidence that Levandowski gave secrets to Uber in order to charge him with a crime. “If I steal your wallet, but don’t spend the money, it’s still a crime,” the official said.

Lawyers for the government say that some trade-secret prosecutions have helped combat thievery. Earlier this year, federal attorneys successfully prosecuted Sinovel, a Chinese wind-turbine manufacturer that had stolen trade secrets from an American company, causing it to lose hundreds of millions of dollars. But in cases like Levandowski’s—in which it isn’t clear what, precisely, was stolen or who, if anyone, was hurt—it’s much harder to argue that American commerce has been put at risk. Nevertheless, Parrella and his colleagues have, with encouragement from Google and Waymo, continued building a criminal case. Insiders say that they aren’t sure yet whether the authorities will bring an indictment, but if Levandowski is arrested it will be the most prominent federal trade-secret criminal prosecution in Silicon Valley history. “Nothing sends a message like a criminal trade-secret case,” the federal official said. “It would go through Silicon Valley like ringing a bell.”

That bell’s reverberations, however, could also disturb what had made Silicon Valley so successful in the first place. Daniel Olmos, an attorney who has represented technology workers accused of trade-secret theft, told me, “You don’t need a lot of cases like this to create a paranoia that stops mobility. You just need one or two high-profile lawsuits.” He went on, “I get calls all the time from scared engineers, who once put some work stuff on their home computer so they could work on it after dinner, and now they’re worried if they try to jump to another firm they’re gonna get sued. And you know what? They’re right to be worried. If a company looks closely enough at anyone’s digital history, they’ll find something they can use against them—some thumb drive they inserted into a computer, some file they e-mailed to themself. And so the easiest decision is usually to ignore the recruiter’s call.”

Some people who have left Waymo told me that they have received frightening letters from their former employer warning them about the dangers of using misappropriated technology—and asserting that Waymo will sue them, if necessary. A former Waymo employee who is now working for another company said that the scare tactics had been effective: “We made a deliberate decision not to do anything that might compete in any way with Waymo. It’s terrifying to think they might come after you.”

Indeed, even if the criminal investigation and the arbitration against Levandowski come to naught, in many ways Waymo and Google have already prevailed. “The people at Google got what they wanted,” one of the lawyers who represented Uber told me. “They got Anthony fired, they distracted Uber and slowed its progress for an entire year, and they let everyone know that if you leave with some of their stuff they can screw with you so bad that everyone will think you’re toxic.”

Although Uber is still developing autonomous cars, many of the employees who joined the company as part of the Ottomotto acquisition have left. Waymo’s development of autonomous vehicles, meanwhile, has sped forward, and its vehicles have now driven more than eight million autonomous miles. In Arizona, the company is rolling out a commercial automated-driving service that will compete with Uber. Larry Page, whom U.S. senators recently criticized as “arrogant” after he refused to testify at a hearing about election security, remains atop the Google empire, and is worth an estimated fifty billion dollars. (Page declined to be interviewed for this article.)

In a statement, Waymo argued that it had sued Uber because “sitting on our hands was not an option to protect the work of hundreds of engineers.” The fact that Judge Alsup asked federal authorities to examine Levandowski’s actions, the statement said, “confirmed the gravity of the allegations and the evidence we had accumulated.” Waymo continued, “We didn’t make the decision to sue lightly,” and added, “Waymo and Google have a long track record of supporting current and former employees in their new ventures.” Although the settlement with Uber represented a fraction of the original demand for $1.85 billion, “Waymo’s primary goal—which we achieved—was to secure protections against the unauthorized use of our technology,” the company said. Waymo declined to answer questions regarding its coöperation with the U.S. Attorneys office, or the company’s policies on providing information to federal authorities, but said, “We comply with law-enforcement requests where there is a valid legal process, and this case is no exception. Waymo’s interactions with the U.S. Attorney’s office are independent from the strength of Waymo’s civil case against Uber.”

Levandowski, for his part, has been out of work since he was fired by Uber. It’s hard to feel much sympathy for him, though. He’s still extremely wealthy. He left Google with files that nearly everyone agrees he should not have walked off with, even if there is widespread disagreement about how much they’re worth. Levandowski seemed constantly ready to abandon his teammates and threaten defection, often while working on an angle to enrich himself. He is a brilliant mercenary, a visionary opportunist, a man seemingly without loyalty. He has helped build a technology that might transform how the world functions, and he seems inclined to personally profit from that transformation as much as possible. In other words, he is an exemplar of Silicon Valley ethics.

Levandowski is upset that some people have cast him as the bad guy. “I reject the notion that I did something unethical,” he said. “Was I trying to compete with them? Sure.” But, he added, “I’m not a thief, and I’m not dishonest.” Other parents sometimes shun him when he drops his kids off at school, and he has grown tired of people taking photographs of him when he walks through airports. But he is confident that his notoriety will subside. Although he no longer owns the technology that he brought to Google and Uber, plenty of valuable information remains inside his head, and he has a lot of new ideas. An investment fund recently started due diligence on one of his proposals: a new self-driving-truck company. He anticipates that some of the funding for it will come from overseas, including from Chinese investors. It is ironic, given that federal trade-secret laws were written to prevent intellectual property from travelling abroad, that a trade-secret prosecution may push Levandowski into foreign hands. But he’s fine with it; what he cares about is having a next act. There are work-arounds, it seems, for everything, even for an unsavory past.

“The only thing that matters is the future,” he told me after the civil trial was settled. “I don’t even know why we study history. It’s entertaining, I guess—the dinosaurs and the Neanderthals and the Industrial Revolution, and stuff like that. But what already happened doesn’t really matter. You don’t need to know that history to build on what they made. In technology, all that matters is tomorrow.” ♦