Cruise COO Daniel Kan (left) and CEO Kyle Vogt (right) in a Chevrolet self-driving-car prototype with a laser sensor on its roof. Kan and Vogt are now senior directors at General Motors, but Cruise retains considerable independence—part of an effort to preserve its fast-moving startup culture.
PHOTOGRAPH BY ART STREIBER

Driven in the Valley: The Startup Founders Fueling GM’s Future

Cruise Automation sold to GM for $1 billion; now its founders are racing to bring a self-driving car to the masses.

Statistically speaking, most of us will never build a billion-dollar company. We definitely won’t build two of them. If any of us did pull that off, we’d most likely feel pretty satisfied with ourselves. We’d retire. Travel the world, maybe. Do a TED Talk, definitely.

Most of us are not like Kyle Vogt, who at the age of 31 has already built two startups that sold for $1 billion. He and Daniel Kan, his cofounder and COO at Cruise Automation, have recently become the youngest senior directors at General Motors, a role most auto-industry lifers would have to toil for three decades to get a shot at. Vogt and Kan got there in just three years because Vogt built self-driving car software so compelling that GM gm had to have it. In May the country’s largest automaker shelled out $1 billion (including incentives for performance) for Cruise, a 40-person startup, before it even launched a product. And before Cruise, Vogt cofounded Justin.tv, a pioneering live-streaming business that eventually evolved into Twitch, a videogame-streaming site that was sold to Amazon amzn for $970 million in 2014.

Electric-powered Chevrolet Bolts in Cruise Automation’s San Francisco garage. Cruise is working with GM to add autonomous-driving technology to the Bolt.PHOTOGRAPH BY ART STREIBER

One of those accomplishments might be enough for a normal human being, but Vogt operates in Silicon Valley, the land of insatiable ambition. Success by normal standards wasn’t enough for Elon Musk or Steve Jobs or Larry Page. And it isn’t enough for Vogt, Kan, and the guys (yes, all guys) in their tight-knit crew of equally accomplished, equally young friends from Justin.tv. Vogt won’t feel like a success until Cruise fulfills its mission: making self-driving cars a reality. Not just a prototype or a few cars in a test market, but fully autonomous vehicles, out in the world, by the millions, in a way that increases access to transportation, reduces congestion and pollution, and, most important, saves lives. To put it in Silicon Valley–speak: He wants to make the world a better place.

Daniel Kan and Kyle Vogt are number 7 on our 2016 40 Under 40 list. See the full list here.

Sitting in a conference room at Cruise’s warehouse-style office in San Francisco’s SoMa neighborhood, Kan, sporting the startup uniform of jeans, T-shirt, and hoodie, matches that Valley platitude with another. “I’ve always heard, ‘The first time you start a company, you want to get rich. The second time you start a company, you want to build a legacy.’ ”

“That’s going to be quoted,” Vogt warns from across the table. (Yep!) Vogt quickly clarifies that Cruise isn’t Kan’s first startup either. “It’s not the get-rich one.”

The startup game has made millionaires out of the people in Vogt and Kan’s network of friends and colleagues, but they’d never say they’re in it for the money. “Your startup should be what you want the first line in your Wikipedia entry to be,” says Michael Seibel, a Justin.tv cofounder who is now CEO of the famous startup accelerator program at Y Combinator (YC). “It’s your life’s work.”

Vogt, a reserved, media-shy Midwesterner, does not have a Wikipedia page yet, but self-driving cars are clearly his life’s work. He’s been passionate about them since participating in the 2004 DARPA Grand Challenge as an undergrad at MIT. At Cruise’s office, he rattles off the ways self-driving cars will improve society, alongside auto fatality stats (35,000 deaths last year in the U.S.). Now, under the umbrella of GM, his tiny software company has a chance to vastly expand its impact.

cru_graphics_2520x1844

But to cement his legacy, he must pull off an even rarer feat—a startup acquisition that actually works. Bright-burning startups tend to fizzle after selling to larger companies. Vogt and Kan must figure out how to navigate the politics and red tape of their new corporate overlord without losing any of the fast-moving startup DNA that made them so attractive to GM to begin with.

The stakes are higher than ever as Google, Apple aapl , Uber, Tesla tsla , Ford, and myriad startups sprint to bring the first self-driving car to market. The resources of a car-building leviathan that’s in dire need of software expertise could give Vogt and Kan the fuel they need to win that race—or weigh them down with baggage that turns them into an also-ran.

cru-10-01-16-bar-comp

Kyle Vogt planned to spend the evening of March 18, 2007, putting the finishing touches on Justin.tv, the live-streaming video site he had cofounded. But at midnight, TechCrunch accidentally published a story announcing the site’s launch, nine hours before it was supposed to go up. An audience of hundreds, maybe thousands, of people flooded onto the not-yet-ready site, causing it to crash. Vogt, the company’s head of engineering, and Emmett Shear, its chief technology officer, “grinded through the night and for two more days,” scrambling to get their site to function properly, says Seibel, a third cofounder. Shear has no memory of the event beyond a photo of himself typing furiously on a laptop from his bed. “It’s all a giant blank spot,” he says. “I hear that happens sometimes with trauma.”

Justin.tv’s goal was to build the first network of live-streaming broadcasters. That sounds a lot less complicated now than it was in 2007, when iPhones were brand-new, bandwidth was expensive, and online video streaming was glitchy. To do what live-streamers can do today with a smartphone, the site’s early broadcasters had to wear elaborate backpacks and baseball hats rigged with camera systems designed by Vogt. Justin Kan, the site’s namesake (and older brother of Cruise’s Daniel Kan), was the site’s first “lifecaster,” capturing himself doing 23-year-old guy stuff: hanging out, wasting time, going to parties, meeting girls.

That chaotic, sleepless first 72 hours foreshadowed the entire seven years of Justin.tv’s existence. Startup life is prone to drama and dysfunction, but Justin.tv’s flailings were so uniquely turbulent that a Hollywood production company optioned the rights for a feature film about it.

cru_graphics_2520x43601

There was the first monthly bandwidth bill for $20,000, double the startup’s cash in the bank. There was the time Seibel had to defend the company at a House Judiciary Committee hearing because users were sharing pirated sports content. The site crashed anytime the Jonas Brothers tried to go live. A teenager live-streamed his suicide. And most famously, in a prank by viewers, a SWAT team kicked down the door of the two-bedroom apartment where Justin.tv’s four founders lived and worked. (The North Beach apartment tower was known as the Y-Scraper because so many Y Combinator founders lived there.)

If Kan was Justin.tv’s star, Seibel was the business guy, Shear was the technical guy, and Vogt was the creative genius. (And for a summer, Daniel Kan was the intern.) Vogt became known for “hero coding” the company out of problems, Justin Kan says. “He’d just, like, lock himself in a room for three days and code away and then emerge with something that worked.”

No amount of hero coding could make Justin.tv into a viable business; 24-hour life­casting wasn’t compelling enough for a mass audience. But while most failed startups are quickly forgotten, Justin.tv’s legacy is traceable throughout the Valley. Young entrepreneurs revel in the mythology of the “Justin.tv mafia,” in part because Seibel and Justin Kan are now leaders at YC. As thirtysomething startup sages, they dispense advice from their time in the trenches to dozens of hopeful founders who come to YC for mentorship, introductions to investors and powerful alumni, and a small seed investment.

And unlike most of Silicon Valley’s graveyard of dead startups, Justin.tv produced a return for its investors. When the original site began to stumble, Seibel spun out Socialcam, a side project focused on short, Instagram-like videos, eventually selling it to software giant Autodesk in 2012 for $60 million. Meanwhile, Shear shifted the remaining team’s focus to videogames, one area of Justin.tv where traffic was booming. Branded Twitch.tv, the site quickly ballooned to 55 million users who spent an average of 20 hours a week watching other people play videogames. Twitch even turned a profit on ads and monthly subscriptions. In 2014, Amazon acquired Twitch for $970 million including incentives—10 times the company’s private market valuation at the time. YC has funded 10 companies that rose to billion-dollar valuations, including Airbnb and Dropbox, but Twitch and Cruise are its two largest exits to date—which just reinforces the Justin.tv mafia’s legacy.

cru_graphics_2520x11801

So many of Justin.tv founders’ Y-Scraper pals became successful that Seibel once said to Justin Kan (in Kan’s words), “Man, we knew so many geniuses.” In reality, timing and grit played a bigger role than IQ. “We’re not like Mozart or something,” Kan says. “We were just in the right place at the right time and we didn’t give up.”

Nor did they dream small. Recently Kan’s cousin, a professor, asked him if he wanted to retire. After all, he’s Twitch-Rich. He told her, “I want to build a billion-dollar company. I want to build a $10 billion company.” The answer surprised his cousin. “I’ve never heard anyone say that,” she said. But though it often feels as if everyone in the Bay Area is swimming in startup cash, the money is beside the point. What matters is what the money symbolizes: Your outsize, possibly irrational ambitions. Kan repeats what I’ve heard from just about everyone in the Justin.tv crew: Every person is the average of their five closest friends. People come to Silicon Valley to be around ambition, he explains. Wanting to build a $10 billion company? Kan throws up his hands. “I’ve only heard people say that.”

cru-10-01-16-bar-comp

The lesson from Justin.tv, according to the founders of Justin.tv, is don’t repeat the mistakes of Justin.tv. And yet, on March 24, 2014, Vogt found himself pulling another last-minute almost-all-nighter to save his new startup, Cruise Automation, from a total meltdown. This time it was literal—the polylactic acid parts he had 3D-printed for his prototype car had melted in the California sun after the first day of YC’s high-profile “demo day” event.

Demo days are like beauty pageants for startups, where founders use big numbers, snazzy pitch decks, and a healthy dollop of snake oil to coax seven-figure checks out of an audience of influential investors. Now Cruise’s star player—Vogt’s black Audi S4, rigged with sensors that enabled it to steer and brake on its own—was a dud. Vogt and his two engineers, Ian Rust and Rita Ciaravino, worked through the night reprinting the parts (they had to buy three new printers to get the job done in time) and rebuilding the prototype from scratch.

The next morning Vogt, looking just barely presentable, drove Justin Kan to day two of the event. For a few miles on Highway 101, Vogt let the car take over. Kan had already invested $25,000 in Cruise “because it’s Kyle,” he says; he immediately invested more. He couldn’t believe they had built a car that drives itself in just three months (and then again, overnight). “Kyle had basically hero coded it once again,” he says.

The Cruise automation team in the garage at their San Francisco headquarters. Cruise’s staff has doubled since the GM acquisition closed, adding person power (and headaches) for Vogt and Kan.PHOTOGRAPH BY ART STREIBER

Cruise emerged from demo day with $4.3 million in convertible debt funding. At the time, Cruise planned to sell $10,000 kits that would retrofit any car to drive itself on highways, similar to Tesla’s autopilot feature today. Daniel Kan, who joined the company just after demo day to run operations, remembers thinking, “This might be one of the craziest things, but it will at least be a good story.” But not long after the demo day, the founders recognized a major roadblock: Cruise would need to customize its kits to accommodate hundreds of vehicle makes and models. In 2015, Cruise abandoned the kits to build software for fully autonomous vehicles.

By then, self-driving cars were feeling less like one of the craziest things. Google was showing off its fleet of jelly bean pod cars, Uber’s CEO predicted an end to human cabdrivers, and Detroit automakers were taking a closer look at the up-and-coming self-driving-car startups.

It’s common for large corporations to get friendly with potential competitors at vague “business development” meetings. Cruise used the meetings with Fortune 500 automakers and suppliers to set ambitious development targets for itself. Startups with no demanding customers can lose their urgency, with development cycles dragging on for years, says Rust, who worked at Google’s experimental research factory before joining Cruise. “We tried to accomplish what one of the most talented teams at Google did in two years and compress that into six months,” Vogt says.

One of the stalwarts, GM, was visiting so frequently that Cruise had to expand its city road testing beyond a single fixed route in San Francisco. “You do that once, and the next time GM comes back, you can’t demo that route again,” Rust says. “You don’t want them thinking, Does this car run on rails?”

“We used it as internal motivation,” Daniel Kan says. “If we impress them, there is potential opportunity; if we don’t, who cares?”

They impressed them. “Every time we went there they’d moved along another nine steps,” says Dan Ammann, president of GM. “We were super excited with what the guys there had achieved already technically, but also the caliber of the talent and speed of development.” In March, GM and Cruise announced the deal, which closed after Cruise settled a spat with an estranged cofounder.

The billion-dollar price tag alone makes the deal seem like a no-brainer for Cruise—but remember, this is the legacy startup, not the get-rich one. Vogt believes selling to GM will help Cruise fulfill its lifesaving mission faster. “If we truly want to do what brought us to work on autonomous vehicles in the first place, we need to do something that is going to achieve scale,” he says. “With GM we can build and deploy it at scale. We want to have the highest possible social impact.”

cru-10-01-16-bar-comp

It’s become a cliché, when a startup sells to a large corporation, for the founders to spout hopeful phrases about “perfect alignment” and “inspiring commitments.” The company is the founders’ baby, and they’ve barely grasped the reality of letting it go. They take the team out for a big celebratory dinner as the congratulations pile up in their smartphone notifications. They take their first vacation in years. They tell themselves the commitments really are inspiring.

It’s also a startup cliché that within a year, the jig is up. The founders realize that the grinding, slogging, against-all-odds hero’s journey of cold calling and pitching and building and hustling and eating glass and staring into the abyss is finally over. Their metabolism doesn’t adapt well to the politics of slow-moving, risk-averse corporations. Once their life’s work begins to feel like a job, a switch goes off in their brains. Some leave to start their next company. Others “vest in peace.” Whatever innovative thing they built gets lost inside a giant corporate overlord. Startups have a 90% failure rate, according to studies. The failure rate for mergers and acquisitions—at least when it comes to meeting expectations—is just as high.

Already Vogt and Kan are being pulled into more tedious corporate meetings, where committees of executives are required to sign off on decisions and budgets that they’re used to handling themselves. It won’t be simple to combine Cruise, a three-year-old upstart that has yet to launch a product, with General Motors, a 108-year-old, 216,000-­person company that shipped 9.8 million vehicles last year to the tune of $152.4 billion. The stakes are high: GM can’t afford to fall behind as the auto industry transforms itself from the business of manufacturing and selling cars to the business of “mobility.”

Vogt and Kan do not intend to become clichés. While negotiating the deal, Vogt turned to his former cofounder, Emmett Shear, for advice. Twitch has thrived under Amazon’s ownership—doubling its users with very little executive turnover—­because it has maintained its independence, with Shear staying on as CEO of Twitch. “[Shear] made it clear to Amazon that it was important for it to be this way if you want to keep the people that made it what it is,” Vogt says. He mimicked Twitch’s deal structure: Cruise operates as an independent subsidiary of GM, and its executives are incentivized to stick around. Cruise’s $1 billion price tag included just $300 million in cash, with another $300 million in GM stock and the rest hinging on Cruise’s ability to retain key employees—mainly Kan and Vogt—and hit technological milestones. “By design, we wanted it to succeed post-acquisition,” Vogt says.

Cruise COO Daniel Kan (left) and CEO Kyle Vogt (right) in a Chevrolet self-driving-car prototype with a laser sensor on its roof. Cruise COO Daniel Kan (left) and CEO Kyle Vogt (right) in a Chevrolet self-driving-car prototype with a laser sensor on its roof.PHOTOGRAPH BY ART STREIBER

Still, there’s a natural culture clash between Cruise’s focus on speed and GM’s focus on durability. Doug Parks, GM’s vice president for autonomous technology and vehicle execution, and Kan and Vogt’s new boss, calls it a “give-and-take.” “They’re focused on getting two to three cars to work, and we’re focused on getting 20,000 to 30,000 cars to work,” he says. “They don’t have expertise on what it takes to make a vehicle durable over many miles.” (The Cruise guys won’t argue there. Vogt’s first assembly-plant tour was “jaw-dropping in its scale,” he says.)

Likewise, GM doesn’t have much experience in the art of “move fast and break things,” but it’s starting to learn. Since acquiring Cruise, Parks’ team has ditched its daily staff meeting where everyone reviews projects, in favor of a Cruise-style daily 15-minute phone call to discuss the biggest issue holding the team back. “It doesn’t allow people to say, ‘We’re going to go off for a week and work on a proposal,’ ” Parks says. “Now it’s, ‘Take the next 23 hours and answer the question.’ ”

Meanwhile, Cruise gets to enjoy the perks of a large, powerful parent company. The startup has access to GM’s testing and validation facilities in Warren, Mich. When something breaks on one of Cruise’s modified Chevy Bolts, Cruise staffers can call the GM team that designed that exact part, rather than spend hours trying to figure out what went wrong. And GM’s brand carries weight. “Suppliers pay attention to us,” Kan says. “You email from your GM email, and it opens all the doors.”

Since the deal closed, Cruise has more than doubled its staff. The company’s office is filling up quickly, something that makes Kan nervous. Office space in San Francisco is already scarce and pricey, and Cruise has a unique challenge: It needs to drive cars in and out of the office. Beneath Cruise’s barnlike warehouse is a fleet of Chevy Bolts with names emblazoned on their hoods (Platypus, Narwhal, Charlie, after Vogt’s dog), each outfitted with lidar laser sensors that look like old-timey sirens. There’s only one ramp in and out of the basement. The cars are starting to feel crowded too.

Cruise could solve its problem by moving its operations to the suburban sprawl of Silicon Valley proper. Like all the automakers, GM has an innovation outpost in Palo Alto. But no one at Cruise has even visited it. Staying independent is too important—and besides, they’re busy. Vogt and Kan can barely spare an hour for a reporter. As I scan my list of questions, Vogt is antsy to get back to his desk. “Is that good for you?” he asks. “I think we’ve given you quite a bit of color.” The clock is ticking, and their legacy is on the line.

A version of this article appears in the October 1, 2016 issue of Fortune with the headline “Driven in the Valley.”

Sign In

Get

Thank you for your interest in licensing Fortune content. Please find information on various licensing contacts below and choose the one that best suits your needs:

  • 1. To license Fortune articles, excerpts, or headlines for republication in various media (including books, eBooks, film, web, newsletters, newspapers, magazines and others), please email syndication@timeinc.com.
  • 2. To license a Fortune cover, order reprint or e-print copies of an article or cover, or license an accolade, please contact PARS International at www.timeincreprints.com.
  • 3. To license text only photocopies of Fortunearticles as print or digital handouts in academic settings, or in academic coursepacks, please contact the Copyright Clearance Center at www.copyright.com