Inside Google's Secret Lab

Google X’s Silicon Valley nerd heaven—America’s last great corporate research lab
Sergey Brin wearing Google Glass Photograph by David Paul Morris/Bloomberg

How this week's cover got made

Last February, Astro Teller, the director of Google’s secretive research lab, Google X, went to seek approval from Chief Executive Officer Larry Page for an unlikely acquisition. Teller was proposing that Google buy Makani Power, a startup that develops wind turbines mounted on unmanned, fixed-wing aircraft tethered to the ground like a kite. The startup, Teller told Page, was seeing promising results, and, he added proudly, its prototypes had survived all recent tests intact.

Page approved Google X’s acquisition of Makani, which was being completed for an undisclosed amount at press time. He also had a demand. “He said we could have the budget and the people to go do this,” Teller says, “but that we had to make sure to crash at least five of the devices in the near future.”

As the polymath engineers and scientists who work there are fond of saying, Google X is the search giant’s factory for moonshots, those million-to-one scientific bets that require generous amounts of capital, massive leaps of faith, and a willingness to break things. Google X (the official spelling is Google [x]) is home to the self-driving car initiative and the Internet-connected eyeglasses, Google Glass, among other improbable projects.

May 22 (Bloomberg) -- Bloomberg's Brad Stone takes us inside Google’s secretive research lab, Google X. He speaks with Emily Chang on Bloomberg Television's "Bloomberg West." (Source: Bloomberg)

The biggest moonshot of all may be the skunk works itself: With X, Google has created a laboratory whose mandate is to come up with technologies that sound more like plot contrivances from Star Trek than products that might satisfy the short-term demands of Google’s shareholders. “Google X is very consciously looking at things that Google in its right mind wouldn’t do,” says Richard DeVaul, a “rapid evaluator” at the lab. “They built the rocket pad far away from the widget factory, so if the rocket blows up, it’s hopefully not disrupting the core business.”

Since its creation in 2010, Google has kept X largely hidden from view. Over the past month, Bloomberg Businessweek spoke to many of X’s managers and project leaders, who work with abundant resources and few of the constraints that smothered similar corporate research efforts in the past. “Anything which is a huge problem for humanity we’ll sign up for, if we can find a way to fix it,” Teller says.

“Anything which is a huge problem for humanity we’ll sign up for”
Illustration by Rami Neimi

Google X seeks to be an heir to the classic research labs, such as the Manhattan Project, which created the first atomic bomb, and Bletchley Park, where code breakers cracked German ciphers and gave birth to modern cryptography. After the war, the spirit of these efforts was captured in pastoral corporate settings: AT&T’s Bell Labs and Xerox PARC, for example, became synonymous with breakthroughs (the transistor and the personal computer among them) and the inability of each company to capitalize on them.

That was last century. NASA’s budget has been clipped by 11 percent since 1990. Companies are pulling back on basic research as well, preferring to buy disruptive innovation when they see it in startups. “I’m pessimistic,” says John Seely Brown, the former director of PARC. “It’s shocking how much research is no longer being done. We have no understanding of how fast China is catching up. I think we are a very complacent nation.”

Google X occupies a pair of otherwise ordinary two-story red-brick buildings about a half-mile from Google’s main campus. There’s a burbling fountain out front and rows of company-issued bikes, which employees use to shuttle to the main campus. Inside one of the buildings, frosted glass covers the conference room windows. A race car tricked out with self-driving technology is parked in the lobby. The car doesn’t actually work; it was put there as an April Fools’ joke. Some of the hallway whiteboards are filled with diagrams of that multigenerational nerd fantasy: space elevators. Media outlets have speculated that Google X is working on such contraptions, which would involve giant cables that connect the earth to orbiting space platforms. Google X is working on no such project, but employees have embraced the concept. It keeps everyone guessing.

Sitting in the passenger seat of a Google driverless car is a test of faith. The car, a white Lexus RX450h with a $65,000 laser range finder on the roof, is cruising at 55 miles per hour on Silicon Valley’s crowded 101 freeway when a giant bus passes—as it happens, a double-decker Google bus, ferrying employees home. As the car weaves to get out of the way, Chris Urmson, the head of the autonomous cars project, is unperturbed. “Google believes in and enables us to do things that wouldn’t be possible in academia,” says Urmson, a former assistant research professor at Carnegie Mellon, his hands resting comfortably in his lap. Google co-founders Page and Sergey Brin “have this idea that incremental improvements are not good enough. The standard for success is whether we can get these into the world and do audacious things.”

“Incremental improvements are not good enough”
Illustrations by Rami Neimi

Last year, Brin, Google’s director of special projects, predicted his company’s self-driving cars will be on the market in five years. Urmson nervously calls that deadline “exciting” and reveals his own target. “I have a 9-year-old son who gets his driver’s license in seven years,” he says. “So I have to be better than that.”

If it weren’t for the robo-cars, there might be no such thing as Google X. The lab’s origins reach back to 2005, when Page first met the Stanford computer scientist Sebastian Thrun at the Darpa Grand Challenge, where Thrun’s team of graduate students was competing to send an autonomous vehicle through a 7-mile obstacle course in the Mojave Desert. The two men shared a belief in the promise of artificial intelligence and robotics and became friends. Two years later, Page convinced Thrun and several of his students to help with its Street View mapping project.

Sebastian Thrun, co-founder of Google X
Photograph by Ian Allen for Bloomberg Businessweek

Thrun had grown disenchanted with the pace of academia, where professors are motivated to publish papers rather than build products. He started the self-driving car project at Google in early 2009. Page and Brin gave him a target: Build one that could flawlessly drive 1,000 miles of open California highways and serpentine city streets. Thrun and his team of a dozen engineers met that goal in 15 months. Their car successfully navigated the jammed streets of Los Angeles and Silicon Valley, and the lower span of the San Francisco-Oakland Bay Bridge, where the car had no GPS reception.

As progress exceeded their expectations, Thrun, Brin, and Page began to talk about expanding the project into a full-fledged research lab. For Page and Brin, it was a way to indulge their longtime interest in technologies beyond search—which generated $44 billion in revenue last year—while keeping the perennially restless Thrun in the fold. “The Google founders were deeply impressed with Sebastian’s ability to be a great scientist who also gets stuff done,” says Teller, a contemporary of Thrun’s at Carnegie Mellon, who joined Google X from a hedge fund, Cerebellum Capital. “Google X was to some extent created as a home for self-driving cars, and literally it was an enticement for Sebastian to stay.”

Thrun always thought of corporate labs as playgrounds for lifetime employees who were overly absorbed by the abstractions of pure research. He wanted to focus on research that was at least commercially plausible and let talent come and go as projects evolved. Thrun says he seriously considered calling the new group the Google Research Institute, but that carried exactly the kind of sleepy connotations he was trying to avoid. Google X, he says, was a placeholder, a variable to be filled in later.

Babak Parviz, creator of Google Glass
Photograph by Ian Allen for Bloomberg Businessweek

Brin decreed early on that the new lab would focus most of its energies on creating hardware. The company’s board of directors funded Google X in January 2010. (Google does not disclose the lab’s budget, but its R&D budget was $6.8 billion in 2012, up 79 percent since 2010.) Google Glass was X’s second project. Babak Parviz, an electrical engineering professor at the University of Washington, who was working on wearable computers, caught the attention of Brin and Page with a paper about the possibility of contact lenses with built-in electronics that could project images onto the wearer’s eye. Combining cars and wearable computing in Thrun’s budding laboratory somehow seemed appropriate. The first Google Glass prototype was a 10-pound head-mounted display with multiple cables snaking down to a box attached to the wearer’s belt.

The latest incarnation of Glass weighs about the same as a normal pair of glasses and is considerably more discreet. The device, which is currently available only to developers, costs $1,500, hangs the equivalent of an HD display over the right eye, and is capable of taking photos and video, displaying e-mail, and subjecting its owner to ridicule. Critics have piled on about Google Glass’s dorkiness—it’s been called a Segway for your face—and, more seriously, its potential for surreptitious surveillance. Parviz wants the world to see Glass in the context of Google X: It’s aimed at making access to knowledge so fast and seamless that it “fundamentally changes the meaning of knowing things.”

Teller compares Glass to the first Apple personal computer. “We are proposing that there is value in a totally new product category and a totally new set of questions,” he says. “Just like the Apple II proposed, Would you reasonably want a computer in your home if you weren’t an accountant or professional? That is the question Glass is asking, and I hope in the end that is how it will be judged.”

Although still young, Google X already has a tradition. As a project evolves beyond the scope of the lab or fails to pan out, the researchers gather for a formal graduation ceremony, complete with diplomas and mortarboards emblazoned with the letter X, to bid goodbye to the people who worked on the project. Last year a team had a graduation after it finished building a neural network that runs on thousands of computers which could learn by surfing the Web.

Astro Teller, Google X’s “captain of moonshots”
Photograph by Levon Biss for Bloomberg Businessweek

Thrun himself graduated from Google X in 2012. He left to form the startup Udacity, which is bringing college courses online. Thrun says Udacity’s focus is “much more of a people problem than a technology problem, and it didn’t fit into Google X.” While Thrun retains an advisory role at the lab, Teller took over as its primary manager, reporting directly to Brin.

In its early years, Thrun’s intuition and Page’s and Brin’s interests guided the agenda. It was Teller who first articulated X’s mission, in a conversation with Page. Teller was trying to put more definition behind the lab’s purpose and asked Page, Were they a research center? No, that sounded boring, Page responded. Were they an incubator for new companies? No.

Finally Teller asked, “Are we taking moonshots?” Page replied, “Yes, that’s it.”

Teller says things like, “We are serious as a heart attack about making the world a better place” with a straight face. He compares Google X to Willy Wonka’s chocolate factory, a magical workshop that needs to be insulated from the world’s judgmental eyes. He recently printed up yellow and green “Save the Oompa Loompas” stickers, which are popular around the lab.

Teller got the nickname “Astro” (real name: Eric) when his high school soccer teammates suggested his spiky haircut looked like AstroTurf. Now 42, he has a shoulder-length ponytail, a salt-and-pepper goatee, and a remarkable, intellectual pedigree. His maternal grandfather, Gerard Debreu, was a Nobel Prize-winning economist; his paternal grandfather, the famed and controversial physicist Edward Teller, worked on the Manhattan Project and is considered the father of the hydrogen bomb (as well as a model for Dr. Strangelove). Astro was close to Edward, who died in 2003. Edward’s troubles after the war, when he was ostracized from the scientific community for testifying against J. Robert Oppenheimer in a security clearance hearing, “is a reminder to me to make sure you do what you love,” Teller says. His grandfather “spent most of his time being a scientific politician and an organization builder because he believed it was critical for democracy and humanity. But he really wished he could have gone in the corner and been a scientist.”

Megan Smith, VP of Google X and co-creator of Solve for X
Photograph by Ian Allen for Bloomberg Businessweek

Teller has turned his sky’s-the-limit thinking into Google X’s most visible export. Last March he spoke at the South By Southwest Interactive conference in Austin, Tex., telling a packed auditorium, “The world is not limited by IQ. We are all limited by bravery and creativity.” Last year, with longtime Google executive Megan Smith, he co-founded the company’s annual, invitation-only conference, Solve for X, a two-and-a-half-day gathering of a hundred or so big thinkers. At the recent session in February at CordeValle, a golf resort south of San Jose, speakers covered topics such as inflatable robots, eye examinations that can detect the early onset of Alzheimer’s disease, and nuclear fusion reactors. “There is really only one guarantee and that is if we don’t try, nothing is going to happen,” said Charles Chase, a senior program manager for Lockheed Martin’s advanced development program, Skunk Works, who gave the fusion talk.

Absurdity is not a barrier to consideration
Illustration by Rami Neimi

When he’s not evangelizing for Google X, Teller spends much of his time surveying new ideas for the lab. Absurdity is not a barrier to consideration. Teller and colleagues say they’ve spent time contemplating levitation and teleportation. The latter was nixed as an area for further study in part because any unique item that you would want to teleport—a Picasso, say—would have to be completely destroyed before it could be reconstituted on the other end.

Some of the real projects in Google X sound almost as outlandish. Makani Power’s newest airborne turbine prototype, called Wing 7, is a 26-foot-long carbon-fiber contraption with four electricity-generating propellers that flies in circles at altitudes of 800 to 2,000 feet, sending power down a lightweight tether to a base station. “If we’re successful, we can get rid of a huge part of the fossil fuels we use,” says Damon Vander Lind, the startup’s chief engineer. Vander Lind acknowledges it might not work, but: “If you don’t take that chance, and put a decade of your life trying to do it, no progress will get made.”

Then there’s X’s still-secret project to bring Internet access to undeveloped parts of the world. A decade ago, David Grace, a senior research fellow at the University of York, spearheaded a project to mount broadband transmitters on high-altitude balloons, as part of a multicountry initiative backed by the European Commission, called the Capanina Consortium. The initiative never progressed beyond the experimental stage. Grace now says that he has heard that Google is working on such balloon-based broadband technology.

Last month, Google Chairman Eric Schmidt made the surprising pronouncement that “by the end of the decade, everyone on earth will be connected to the Internet.” Skeptics immediately noted that 60 percent of the world is not yet online and that there are many countries without even reliable telecommunications grids. Teller won’t confirm or even discuss such a project, though he concedes that wiring the planet would fall squarely into Google X’s purview. Grace says, “It does need the Googles of the world to push this forward.”

While Teller runs the day-to-day operations at X, he reports to Brin. (“Sergey is Bruce Wayne, and I’m Lucius Fox,” Teller says.) Colleagues say that since Page became CEO in late 2011, Brin spends most of his time immersed in the technical details of several projects at Google X. Although he declined several requests to speak for this story, on a typically bucolic day at the Google campus in Mountain View, Calif., Brin happens by a reporter and two Google spokespeople eating lunch outside and spontaneously joins the group. “I think I’m going to spend most of my time now on cars,” he says, by way of introduction. Asked about Google Glass, a project he championed and which he has been photographed testing on the New York City subway, he points to the device perched on his nose and says, “You know, this is basically done.”

Inside the Moonshot Factory
Illustration by Rami Neimi

Now he’s ready to throw himself into driverless cars. Part of Brin’s charm is that he speaks in flat, logical sentences that are as dull as timber. But when he gets going on a soliloquy, he can build a pretty astonishing house. Autonomous cars may seem like a gimmick, he begins, but when you consider all the time that people won’t be devoting to their rearview mirrors, and all the efficiencies that come from cars that could be zipping between errands rather than idling in parking lots, the world looks like a very different place. Car ownership would be unnecessary, because your car (maybe shared with your neighbors) will act like a taxi that’s summoned when needed. The elderly and the blind could be thoroughly integrated into society. Traffic deaths could be eradicated. Every person could gain lost hours back for working, reading, talking, or searching the Internet. If it all comes to fruition—a big and fascinating if—Brin could be one of the few entrepreneurs to have changed the world twice.

Brin’s presence at the lab is one of Google X’s most powerful assets. “At some other company, you might imagine a senior vice president who feels inevitable pressure to claw back the smartest researchers to go put out a fire elsewhere,” says DeVaul, who helps Teller survey possible new projects. “Sergey’s direct involvement is one of the ways this environment gets supported.” As a result, Google X is both growing in ambition and stockpiling an impressive roster of technical talent.

Mary Lou Jepsen, a former Massachusetts Institute of Technology professor, joined last year as the head of the Google X Display Division, where she is presumably developing screens for the coming wave of devices that, like Glass, will perch on or otherwise cling to our bodies. Andrew Conrad, formerly the chief scientist of LabCorp, the blood-screening firm, joined Google X this spring to work on an undisclosed project. Jeff Huber, the longtime Google executive who most recently ran Google’s mapping and commerce divisions, also moved to X in March but did not specify his area of research.

The expansion of Google X is enough to give some Google investors a touch of heartburn. Brian Wieser, an analyst at the Pivotal Research Group, calls Google X a “benign to positive” factor for shareholders but wonders why the company doesn’t license technology such as Google Glass. “When it comes to creating businesses that are arguably a strategic overreach and likely margin-eroding, that is when you get concerned,” he says. Other investors remember that the founders’ ranging curiosity is what led to winning bets on businesses that seemed irrelevant at the time—Android, for example, which runs on 75 percent of all smartphones shipped around the world in the first quarter, according to the research firm IDC. “It’s a culture that has enabled them to not get caught flat-footed by transitions,” says Nabil Elsheshai, an analyst at Thrivent Financial for Lutherans, which holds Google stock.

Teller says he wants Google X to be judged not only on its financial return but on the progress it makes toward clean energy or wiring the world, or on its other projects. “We are still in our adolescence,” he says. “We are still figuring out how to do things, like how to kill projects or amplify them when we decide they need to go into the next stage.” For now, X will take on two or three new moonshots a year. “If there’s an enormous problem with the world, and we can convince ourselves that over some long but not unreasonable period of time we can make that problem go away, then we don’t need a business plan,” Teller says. “We should be focused on making the world a better place, and once we do that, the money will come back and find us.”

Before it's here, it's on the Bloomberg Terminal.