Alphabet is ending an era of unrestrained invention, spinning off many projects into independent startups.
Skip co-founder Kathryn Zealand holds the first functional prototype of the startup’s wearable tech at its San Francisco labs. Photographer: Gabriela Hasten for Bloomberg Businessweek
By Julia Love
May 20, 2024 at 9:00 PM UTC
If there were an aptitude test to qualify for a job at X, Alphabet Inc.’s renowned research lab, Kathryn Zealand would ace it. She’s a McKinsey & Co.-trained entrepreneur who abandoned a Ph.D. in theoretical physics to build startups in sub-Saharan Africa. In her spare time, she climbs California’s tallest mountains while bingeing sci-fi audiobooks. She joined X in 2018, and for the past five years, she’s dedicated her life to building an exoskeleton.
On a recent afternoon in San Francisco, Zealand donned a prototype of her product, a mechanical system designed to enhance a person’s musculoskeletal abilities. This early version is a belt studded with wires, actuators and sensors that’s not particularly fashionable.
There aren’t many companies out there making exoskeletons, but the few that do mostly target warehouse or factory workers at risk of injury or in need of a boost to lift heavy objects. Zealand is developing a pair of battery-powered trousers to assist people who struggle with mobility. It’s a noble aim and a devilishly difficult design challenge. The device must be lightweight, powerful and stylish enough that people will wear it in public. The prototype is sufficiently light and comfortable, but Zealand is struggling to perfect the fit; the belt gapes at her hips. “A prototype is always built to answer a question,” she says.
The main question she’s trying to answer is whether she can finally realize her vision for the product outside of Alphabet. Last year she and a small team left X’s lab and established an independent startup. They traded a world-class design studio, generous compensation and the prestige of the Google brand for autonomy. The exoskeleton “is almost like a great project for X 10 years ago,” Zealand says. “But by the time we got around to solving all the details, we were less on-brand.”
Skip is creating wearable tech for mobility assistance. Photograph by Gabriela Hasten for Bloomberg Businessweek
Alphabet’s self-proclaimed moonshot factory is coming back down to earth. From its founding in 2010, the lab captured the public’s imagination with visions of a techno-utopian world it would help invent. In the not-so-distant future, X posited, everyday people might ride to work in a self-driving car (Waymo) and socialize with a powerful computer resting on the bridge of their nose (Google Glass). The most whimsical idea was at Project Loon: A network of high-altitude balloons would beam internet to remote locales, creating educational and economic opportunities for previously disconnected communities.
The leaders of the lab readily acknowledged that many of these efforts would not pan out. Yet Google’s willingness to invest in speculative technology sent a powerful message to Wall Street and Silicon Valley, echoing a defiant letter penned by co-founders Larry Page and Sergey Brin when the company went public. (“Google is not a conventional company. We do not intend to become one.”) X’s formation signaled that, even while the namesake business was still growing rapidly, Page and Brin had set their sights on hatching the next Google and solving some of humanity’s thorniest problems.
X enjoyed a privileged place in Mountain View, California, for years, and Brin regularly worked from an office there. But former employees say Google has become less a hothouse of innovation than a cautious custodian of its search empire, which appears to be increasingly under threat. OpenAI’s ChatGPT unleashed a new class of software that can succinctly answer questions and perhaps obviate the need to search on Google. This prospect set off a panic at Google’s headquarters that has transformed the company. When Brin turns up at work now, he’s more likely to be writing code for Google’s AI projects, far from the Xplex, say people with knowledge of the matter.
X’s office in Mountain View. Photographer: Cayce Clifford for Bloomberg Businessweek
To sharpen its focus—a phrase Alphabet Chief Financial Officer Ruth Porat and Chief Executive Officer Sundar Pichai often use on earnings calls—the company began reining in cushy corporate perks and conducting rolling layoffs. Google’s beloved 20% time, in which employees can devote one day a week to passion projects, is technically still on the books, but rank-and-file employees say they have little freedom to indulge in it. With the exception of artificial intelligence, a natural complement to the core business, “I do get a strong sense that Google has long since given up on breaking new ground,” says one person, who, like many of the more than two dozen current or former employees interviewed for this story, requested anonymity for fear of being fired or facing other professional repercussions. “They’re pretty much just turning the crank.”
Top executives seem to see the idea of launching internet balloons into the stratosphere as an unjustifiable expense when the core business could be in peril. But the uncomfortable truth is that ChatGPT was built on a discovery that came partly out of work at X, when AI still sounded like sci-fi. As Alphabet became frozen with fear, OpenAI emerged as an innovation hub, with a research-driven culture instilled by former Googlers. But OpenAI, as its name implies, is only focused on AI. The era of freewheeling invention lavishly funded by a tech giant that X embodied is largely gone. Amazon is culling dozens of projects, Apple gave up on building a car, and Meta shut down some of its hardware efforts. These days, when people talk about X, they’re usually referring to Twitter under Elon Musk.
As for Alphabet’s X, one of the primary goals for projects had always been to earn their place as a standalone business unit of the parent company. (This is partly why Google changed its name to Alphabet in the first place—a company for every letter.) But that process is on hold for at least the next year, say two people with knowledge of the matter. Instead, X is carving out a path in which projects can spin off as startups, like Zealand’s did.
It’s the result of years of debate and experimentation between X and its corporate overlords, current and former employees tell Bloomberg Businessweek. The company was initially reluctant to let outsiders share the fruits of its investments or risk compromising intellectual property, including code that often relied on the work of other teams at Alphabet. But executives ultimately decided it was better than letting promising technology wither on the vine.
“X was created as a ‘Moonshot Factory’ to create new technologies to solve complex problems,” Porat wrote in an emailed statement. “We are excited that X leadership is increasingly pursuing opportunities to scale and monetize many of those innovations to deliver sustainable value creation.”
While the new policy opens up more possibilities, it also signals that Alphabet will be shutting off its spigot of cash to more mature projects that haven’t proven themselves financially viable. X’s budget this year is less than what it was last year, and employees worry about what that means for next year. Some former X employees say the pivot to startup incubation represents a hollowing out of the lab and a retreat from its central purpose of jump-starting the next Google.
Astro Teller, who’s been running X for more than a decade, downplays the significance of the shift. A futurist who reports to work on rollerblades, Teller says that from X’s earliest days, the lab has been trying to balance efficiency with innovation. He believes so deeply in the lab’s model, he says, that he doesn’t have to worry too much about the prospects of any one project.
“We’re hardly perfect. We’re in the middle of our journey somewhere,” Teller says. “But we’re enough better than random that I sleep well at night.”
Teller, the head of X, at his lab. Photographer: Cayce Clifford for Bloomberg Businessweek
Some of Silicon Valley’s biggest breakthroughs came from a company that made copy machines. Xerox’s research-and-development lab in Palo Alto invented ethernet, the graphical user interface and other pillars of personal computing. But Xerox PARC failed to commercialize most of the technologies, allowing Steve Jobs and others to do so first. The rise of Apple Inc. and fall of Xerox Corp. became a parable about big companies failing to recognize the value of their own inventions.
From Google’s early days, Page and Brin made a point of funneling profits from their search engine into businesses that had nothing to do with it. Out of this effort came Gmail and Google News. They institutionalized this philosophy with X and tapped Sebastian Thrun to head it. A leading robotics engineer and Stanford University professor, Thrun studied the legacies of some of the most famous corporate research labs to define X’s key tenets: hire the best minds, set audacious goals, dole out bonuses.
Notably absent from these principles was any spirit of fiscal discipline. If Thrun had a budget, he was only vaguely aware of it. He gave his deputies as much money as they requested, and his only ask was that they not spend it all that year. Thrun reasoned this would inspire restraint, he says, in the same way that employees with unlimited vacation take fewer days off. The sort of efficiency he was more concerned with involved scolding teams that had too many meetings. “The culture we set up was one of complete trust,” Thrun says. “I would do whatever it took to make these teams successful. If they were doing great, they would not see me at all.”
One of the first teams at X, led by another Stanford professor, Andrew Ng, sought to build early AI software inspired by the architecture of the human brain. X gave him the ability to operate at the speed of a startup while drawing from Google’s deep well of talent, Ng says. The parent company eventually snapped up the project, called Brain, and made it a cornerstone of Google’s AI efforts, but the time at X was formative, says Ng, who’s now on the board of Amazon.com Inc.: “I give Google X a ton of credit.”
Thrun’s right-hand man at X was Teller, a scholar-turned-novelist-turned-entrepreneur on wheels. His given name is Eric, but everyone calls him Astro, a nickname he acquired in high school when friends said his hair reminded them of astroturf. His family tree includes a Nobel Prize winner, a philosopher and a complicated historic figure: his grandfather, Edward, the creator of the hydrogen bomb. In a 2018 interview with the Atomic Heritage Foundation, Teller drew parallels between his work at X and the development of the first atomic weapons. “I could easily describe many of the things that we’re doing here as ‘a Manhattan Project for blank,’” Teller said. “Calling them ‘moonshots’ rather than ‘Manhattan Projects’ is probably better marketing.”
When Thrun left X in 2012, he handed the reins to Teller. The early crop of projects were mostly planted by Page and Brin, and Teller wanted new ideas. To systematize the creation of projects, Teller built out a team known as Rapid Evaluation. This group stress-tested dozens of ideas with the intent of surfacing the most promising. A stable of lawyers and accountants was hired to handle issues like regulatory compliance and contracts, so the entrepreneurs didn’t have to trouble themselves with such concerns. As X grew, Teller shielded it from the outside world and even the rest of Google. He believed such isolation was necessary for X to “make a huge impact in the world and also make a huge return for Google,” says Will Patrick, an early member of the Rapid Evaluation team.
But the huge returns were theoretical, and X couldn’t hide from financial statements for long. In 2015, Page and Brin restructured Google as a conglomerate, with X as the breeding ground for new ventures. Employees were at first energized by the shift, which endowed their work with greater purpose. But the process of Alphabetization brought more scrutiny. Investors had never known before exactly how much Google spent on X, because its costs were lumped in with the rest of the company’s. Now shareholders had clearer visibility thanks to a new line item: Other Bets.
Porat, who joined shortly before the formation of Alphabet, began meeting regularly with Teller and scrutinizing projects in X’s portfolio. Around this time, X developed a system designed to estimate the commercial viability of each moonshot, says Loon co-founder Cliff Biffle. The initiative was called Moneyball.
Still, X remained a magnet for talent inside and outside of Alphabet. It was where the board met and the founders spent time. Brin regularly brought Bono and other friends for tours. His compound on the third floor, a wing of conference rooms he shared with Page that required special badge access, was the subject of so much intrigue that it became known as Narnia, say four former employees.
Even under the watchful eye of Porat, X’s budget continued to grow. Colleagues were reluctant to criticize one another’s work, and troubled projects would grow unwieldy until Teller would finally withdraw support, former employees say. Ones that did get the ax were commemorated at Dia, an annual party inspired by Mexico’s Day of the Dead that featured dancers, margaritas and a tequila tasting. The ritual was meant to portray failure as a natural part of a team’s life cycle, a cause for celebration and an opportunity to begin anew.
Ventures that might have succeeded as standalone companies were sometimes absorbed into Google. After starting Loon, Biffle worked on a computer chip customized for machine learning software, which drew interest from prospective customers outside of Alphabet, he says. He and his team were delighted—until Alphabet folded the product into another AI unit of the business around early 2019. “Google is a technology-hoarding company,” says Biffle, who now works at a startup that makes servers.
Teller’s philosophy of isolationism sometimes conflicted with Porat’s goal of profitability. One X project that sought to set up a system to accept payments from outside clients couldn’t do so because it required approval from another part of Alphabet, Biffle says. “People outside X are not incentivized to help X,” he says.
In 2019, Page handed over his role as Alphabet’s CEO to Pichai, the head of Google. Financial scrutiny intensified soon after, and the lab’s budget, which totaled hundreds of millions of dollars, flattened and then declined, former employees say. A spokesperson for X says there had always been some fluctuation in X’s budget based on when projects graduated and moved off the lab’s books. Earning a diploma from X doesn’t guarantee success. In 2021, Alphabet shut down Project Loon, the balloon internet initiative which had long stood as a symbol for X’s lofty ambitions. By that point, the return on investment that executives at Alphabet and X expected of moonshots was gradually rising. At an all-hands meeting in 2022, Teller asked employees to put themselves in Pichai’s shoes and imagine how they’d feel about a $1 billion business. He advised them to aim much higher.
Crafting a business plan for a project like an exoskeleton often felt like writing science fiction, but Alphabet’s leadership was unwilling to spend heavily without first considering a business model, as it had done for Loon. What troubled some employees the most were the aging projects that lingered in the lab, competing for dwindling resources. Quarter after quarter, these projects met their goals by reaching new technical milestones or hitting a revenue target, but they had no apparent prospects of graduating into Alphabet. It prompted some employees to conclude that the only way forward might be out.
The Googleplex, Alphabet’s main campus. Photographer: Brooks Kraft/Corbis/Getty Images
Kathy Hannun joined X shortly after graduating from Stanford and quickly rose the ranks to project leader. But by 2016, a divide between her and Teller was emerging. Hannun had been hard at work on an effort to develop geothermal heating and cooling systems for people’s homes. Geothermal energy offers an environmentally friendly alternative to propane and other fossil fuels, and Hannun was eager to launch a pilot in New York state. But Teller wasn’t ready to start digging up people’s backyards. He told Hannun there could well be a viable business in geothermal heating—but not for Alphabet.
Hannun was crushed. She loved working at X, but she wasn’t prepared to abandon her project. She was determined to continue her work on the outside, and Teller agreed to explore the option. Over the course of nine months and dozens of meetings with X’s leadership and Alphabet’s corporate development team, Hannun arranged a deal in which a newly formed company, Dandelion Energy Inc., would receive the intellectual property underpinning the project, and Alphabet would retain a minority stake. It was X’s first spinout from Alphabet.
Within months of her exit, Hannun and her small team were drilling small holes behind people’s houses and installing geothermal systems. She credits Teller for the opportunity. “Dandelion has been successful because he pushed us out of the nest, and we were forced to figure it out,” she says.
Although Hannun had broken free, a path outside of X wasn’t available to everyone. Former employees say teams at X found it difficult to hold the attention of the corporate development group, which is used to negotiating multibillion-dollar deals for Google. Members of X’s finance department described the process with Hannun’s company as a “distraction,” she recalls. The deal also forced the company to reckon with fundamental questions about why X even exists, says Obi Felten, one of Teller’s former lieutenants, who helped organize the Dandelion spinoff. “If you define the purpose of X very narrowly, the purpose of X is to make new Alphabet companies,” she says. “There was a philosophical question: Is this our job, or is this a waste of resources if they become a startup?”
After Dandelion, Alphabet evaluated each business proposition on its own merits. One unloved X project called Project Malta, which sought to store energy in molten salt, struggled to secure Alphabet’s blessing for its team to spin out. In 2018, Alphabet came around after an ex-member of the Malta project who’d become a venture capitalist agreed to invest, says Ty Jagerson, a former X employee who joined Malta after the spinoff. But efforts to save Loon were unsuccessful. Shortly before the balloons were grounded for good, an outside investor expressed interest in an acquisition, say four former X employees. Alphabet nixed the deal in part because Loon had been built upon layers of internal Google code and tools.
Calls for spinoffs, most of which were still going unanswered, grew louder during the venture capital boom of 2021, when investors were flush with cash and seeking to capitalize on a pandemic tech bubble. X, which for so long had promised the perfect balance of Big Tech stability and startup-style experimentation, was starting to feel to some employees like a trap. And even Felten, who’d helped facilitate some of the first spinoffs as Teller’s deputy, has joined the startup fold. While at X, she worked on a project that aimed to find a biomarker associated with anxiety and depression. After X shut down the project in 2021, she used her personal savings to buy the rights to some of the IP and founded her own company, Flourish Labs.
As for Zealand’s exoskeleton, the project while at X went through a familiar trajectory of great promise followed by disenchantment. Her first prototype withstood rigorous testing, including a dash up 61 flights of stairs to the top of Salesforce Tower, San Francisco’s tallest skyscraper. Teller praised Zealand’s team as “Xey,” a term of endearment that connotes a blend of scientific rigor and whimsy. (It’s pronounced like “sexy” without the “s.”)
But by 2022, Teller and his executive team were encouraging Zealand to move in a different direction. Some favored courting corporate customers looking to increase efficiency in warehouses. Others proposed making immersive video game controllers. Zealand agreed to explore the potential of these markets but was left more convinced that her original focus on human mobility was more promising. On top of that, she was frustrated by obstacles she faced at Alphabet while trying to get approval for partnerships, including one with a major athletic brand, says a person with knowledge of the deal who asked not to be identified because the details are private.
For advice, Zealand quietly contacted former colleagues who’d successfully spun out. She anticipated a protracted negotiation, but within four months, she and her team were out. Financial terms weren’t disclosed, but what’s clear is that life after Alphabet isn’t glamorous. They took pay cuts of more than 50% and work in a small office with a less robust selection of snacks. VCs have expressed concerns about the startup’s unconventional beginnings. “When I would talk to investors, they would be like, ‘How do we know that you’re not the dud? Why doesn’t Alphabet want to keep you?’” Zealand says.
So Teller wrote a letter of recommendation on behalf of Zealand’s startup, Skip Innovations Inc. She’s managed to raise $6 million in total funding, and the company is engaged in a clinical trial of one of its devices involving patients with Parkinson’s disease. (She declined to name the partner.) Zealand hopes to reach a mass market that includes those with milder movement needs. Starting this summer, anyone will be able to preorder one of Skip’s devices.
Zealand appreciates the benefits of independence—like the ability to invite a reporter to her office in San Francisco for a demo. Envisioned as a salve for knee pain, the device I tried was shaped like a thigh-high boot and packed with sensors and motors to study my gait and deliver support at just the right moment. Before I took my first step, I could feel the motors twitching in anticipation. When I began to walk, they kicked into gear, reducing the force my muscles needed to exert; their hum crescendoed to a dull roar when I ascended a staircase.
Near the exit of Skip’s office, there’s a glimpse of the company’s potential future: A mannequin wears a pair of pants molded with motors around the hips, thighs and shins, all fashioned in khaki green, fit for a bionic woman heading to yoga class. The reality, though, is that most startups fail, and if that happens in this case, there will be no Alphabet job security at the end of it. It’s a risk Zealand is at peace with. “I didn’t come to Silicon Valley to stay in a cushy job,” she says. “Life is for living.”
Inside X’s office. Photographer: Cayce Clifford for Bloomberg Businessweek
On a recent morning, Teller glided across the X lobby wearing dark jeans, a gray button-down and black inline skates—business casual on blades. The Xplex, a corridor employees pass through on their way to work, is equal parts museum and mausoleum. In the most prominent exhibit, one of the last remaining Loon balloons rests in a glass case, never again to drift beyond the low-slung ceilings of this converted shopping center.
Teller started to tell a well-worn story of how X set out to solve the world’s connectivity challenges, but then, rather than explaining the team’s failure to ground the soaring cost of the balloons, he whirled around and pointed to another piece of hardware hanging from the rafters: a communications terminal. It had been made by Project Taara, a newer effort to deliver wireless, high-speed internet that had come out of Loon. Instead of using balloons, Taara transmits data through beams of light between two terminals, billed as an alternative to fiber optics in places where laying cable isn’t feasible, such as across rivers and on protected land. It’s seen as one of X’s most promising spinout candidates. “We call it moonshot compost,” Teller says. “Even if we have to come at things two, three, four times, we often end up with really good solutions.”
The tour through Alphabet’s exploratorium touched on agriculture, aquaculture, renewable energy, drone delivery and industrial robotics. The museum’s polish faded by the end, when Teller reached a pair of oversize poster boards papered with printouts showing X spinoffs, including Dandelion and Skip. The space would soon be renovated to illuminate the diverse paths that X companies have cut through the world. Although it took Teller time to appreciate the value of these exits, he says he’s come to embrace them.
“We’ve become clear that actually there’s an opportunity for us to do more good for the world faster by launching things earlier, getting that sort of market-based capital sooner,” Teller says. “That allows us to stay really focused on moonshot origination, which is arguably the thing that we’ve sort of honed and gotten to be the best at.”
As more moonshots prepare to spin out, Teller has assembled a team to assist with what he calls the “midwife process.” Ivo Stivoric, a longtime deputy of his, is the chief midwife. He’s been meeting with prospective investors and coaching moonshot leaders on how to pitch, a skill many never needed until now. Alphabet’s own investment firms CapitalG and GV are unlikely to be among the investors, a spokesperson says, in part because X takes a stake in each spinout.
At least some projects on the spinoff track can still hire, but job postings now include a disclaimer: “If this project is successful, this role will immediately transition into the same full-time role at an exciting early-stage startup outside of X,” according to an ad for Tapestry, an X moonshot that seeks to organize electric grid data. The posting suggests Tapestry will spin out in six to nine months. Page Crahan, its general manager, says she’s excited to have more freedom to work with partners. “Who knows what we’ll be able to do, but it feels like another place that we can be creative,” she says.
Meanwhile, Alphabet seems to be digging through the compost for products that could warrant revival. On May 14, when Google unveiled new computer vision software, Brin mused that Google Glass, the early face computer he championed before it was ridiculed to death, would be “the perfect hardware” for this technology if only it weren’t 10 years too early. Brin continues to keep a watchful eye on X’s work and was among the backers of a carbon capture startup called 280 Earth that spun off in 2022. Gideon Yu, a co-owner of the San Francisco 49ers who also invested in 280 Earth, says he’s eager to fund more startups from the lab, because the ones that survive long enough to spin off are among the “hundreds of ideas that have been winnowed down to the very best.”
X is looking into raising a fund so it can invest in startups that emerge from the lab, according to a person with knowledge of the plan. But the team is still learning what VCs want. One startup investor who reviewed pitch materials for Tidal, an aquaculture moonshot, says Alphabet was seeking an overinflated valuation for a business with such paltry revenue. The VC, who asked not to be identified discussing private company documents, says Alphabet hadn’t even assigned a valuation to its Tapestry deal at that point, and he questions in general the motivations of X-bred entrepreneurs accustomed to Google’s gourmet cafeterias and generous compensation.
Teller says X always intended to prepare employees for survival without X. In the early days, he instituted one of the lab’s first rituals: graduation. When projects were swallowed up by Google or later given a place within Alphabet, workers would gather for a ceremony featuring mortarboards and diplomas. Teller started holding the events, he says, to temper the sadness of leaving X, which had become a coveted perch within Google. “I wanted them to understand, X is not a permanent place for almost anybody,” he says. “I didn’t want people to feel kicked out of the Garden of Eden.”
Graduation ceremonies will continue, but now that some dead X projects can find an afterlife as a spinout, the future of the Dia parties is uncertain. The event wasn’t held last year due to budgetary constraints, and the employee who organized the previous ones was laid off in a round of cuts targeting some of the lab’s veterans.
Teller, meanwhile, maintains his daily rituals, even as Alphabet and the rest of the Valley have changed around him. At 53, he still spends the full workday on rollerblades, though his Achilles tendon bothers him from time to time. “I tell myself it has nothing to do with the rollerblades,” he says. His employee badge, which features a snapshot from a time when his beard was streaked with fewer strands of gray, swishes at his hips as he skates.
Ideas cycle and recycle in Silicon Valley. Even the reinvention of Teller’s lab echoes history. In one of its later chapters, Xerox PARC rebranded as PARC and adopted a business model in which it would partner more closely with the industry and spin out its most promising inventions into independent companies. One of those spinoffs was sold to a three-year-old search engine company run by a young Page and Brin. —With Sarah McBride, Antonia Mufarech and Mark Bergen
Source (Archive)
