How GE Exorcised the Ghost of Jack Welch to Become a 124-Year-Old Startup
Overlooking the Hudson River in Ossining, N.Y., there’s a grassy, 59-acre campus owned by General Electric. It’s an executive training center where the company holds management and leadership classes, some of them led by the chief executive himself. Jack Welch, who ran GE in the 1980s and ’90s, would arrive by helicopter. He’d make his way to a windowless auditorium known as the Pit where a group of managers waited. They used to call him “Neutron Jack,” because he was known for firing so many people that only the buildings were left standing. Neutron Jack and his executives would engage in an aggressive form of corporate group therapy, raising their voices as they aired their frustrations with the company and each other. Later, they would have drinks at the White House, the campus bar. It drove business magazines wild with excitement. “The class sits transfixed as Welch’s laser-blue eyes scan the auditorium … , ” wrote Businessweek in 1998.
Today, GE executives—sorry, team members—take classes in yoga and meditation and suminagashi, the Japanese art of painting on still water. The White House has become a low-key place where visitors can sip artisanal coffee rather than martinis. The Pit has a window through which the sun shines.
It’s part of a much larger transformation at GE orchestrated by Jeff Immelt, Welch’s successor as chief executive officer. Most notably, GE is moving its headquarters from suburban Fairfield, Conn., land of golf and bonuses, where it’s been since 1974, to Boston, the Athens of America. The company is selling off its division that makes refrigerators and microwave ovens. Now it’s focused on electric power generators, jet engines, locomotives, and oil-refining gear. And it’s made a significant bet on developing software to connect these devices to the Internet. There’s a term for this trend of adding network connections to hardware not usually considered computers: the Internet of Things. GE believes its opportunity lies in what it calls the Internet of Really Big Things.
In the past five years, GE has hired hundreds of software developers, created its own operating system, and fashioned dozens of applications that it says will make planes fly more efficiently, extend the life of power generators, and allow trains to run faster. GE’s plan is to sell this software to other manufacturers of Really Big Industrial Things, and to be a top 10 software company by 2020. That would put it in the same category as Microsoft, IBM, and Oracle, an ambition that some have difficulty swallowing. “Top 10? No way,” says David Linthicum, senior vice president of Cloud Technology Partners, a consulting firm in Boston.
GE is also revising its managerial rhetoric, something it’s also historically produced in prodigious quantity. The company was officially founded in 1892 when Thomas Edison merged his operation with a rival electric light manufacturer. In the 1950s, CEO Ralph Cordiner promoted the theory of decentralization, which turned 120 business heads into mini-CEOs. In the 1970s, Reginald Jones championed “strategic business planning,” which treated the company’s many ventures as an investment portfolio. As a sloganeer, no one matched Neutron Jack’s ferocity. He was an evangelist for Six Sigma, a numbers-driven quality-control method that he didn’t originate but grabbed hold of and turned into a boardroom craze. He wanted GE to be a “learning enterprise” with “a boundaryless culture.” He also called it “the greatest people factory in the world,” one that welded together managers who could run anything from the plastics division to a television network. (Welch spent several years dispensing management advice in the pages of this magazine a decade ago, after retiring from GE.)
Immelt, who took over in 2001, tried to promote his own management methods. He brought in cultural anthropologists to study employee behavior. He tried to get his executives to submit “imagination breakthroughs” that would galvanize GE and generate growth. GE held “idea jams” to foster creativity.
Nothing seemed to work. GE’s shares were mauled in the recession of the early Aughts. Meanwhile, its GE Capital division morphed into one of the world’s largest providers of commercial real estate debt and aircraft leases. During the financial crisis of 2008, Immelt was forced to seek the protection of the Federal Deposit Insurance Corporation, which guaranteed about $60 billion of GE Capital’s debt. The same year, after GE missed its quarterly earnings projections, Welch declared during an appearance on CNBC that Immelt had “a credibility issue” and threatened to get a gun and shoot him if he did it again. The Financial Times reported that Immelt complained to a group in Washington that he had the misfortune of managing GE in a turbulent time. “Not only could anyone have run GE in the 1990s,” Immelt groused, “his dog could have run GE. A German shepherd could have run GE.”
By then, Immelt had seen the share price fall from $60 in 2000 below $6. GE was stripped of its triple-A credit rating by Standard & Poor’s, and Immelt cut the dividend for the first time since 1938. In 2012 this magazine referred to Immelt’s first 10 years as “GE’s Lost Decade” and calculated that investors had seen a total return of zero during his tenure.
Inevitably, an activist took an interest in the struggling conglomerate. Last October, Nelson Peltz’s Trian Partners revealed that it had purchased $2.5 billion in GE shares, becoming its ninth-largest investor. In an 81-page analysis, Trian said GE had previously been an unfocused, overly bureaucratic muddle. But rather than call for a breakup of the company as Peltz has done in the past with DuPont and PepsiCo, he instead endorsed Immelt’s strategy.
A year earlier, this would have been hard to believe, but by last fall, Immelt’s program was beginning to succeed. He had announced a plan to shed $200 billion of GE’s problematic financial assets, which have weighed down its share price. The company says it had software sales of $5 billion in 2015, a sign that the Internet-of-really-big-things approach must be taken seriously. And, by all accounts, Immelt’s campaign to remake the company’s intrinsically rigid culture is working. In the past year, GE’s stock has outperformed the Standard & Poor’s 500-stock index. “A lot of people didn’t think this management team would drive an aggressive transformation of the business,” says Steven Winoker, an analyst at Sanford C. Bernstein & Co. “But that’s exactly what’s happening.”
Immelt enters a conference room at GE’s 53rd-floor office in Rockefeller Center in New York. He’s 60, 6 feet 4 inches with wavy white hair, and still exudes youthful confidence and self-deprecating charm. He’s a former Dartmouth football player and fraternity president who once told this magazine that he won the Earl Hamilton Varsity Award for friendship and character, adding that it probably went to the campus beer-drinking champion.
Although he often wears jeans to the office, Immelt has a board meeting later this February morning, so he’s dressed in a light gray suit, pink shirt, and green tie. After some breezy small talk, he starts going into GE’s transformation, which began in the depth of the financial crisis. To hear him tell it, he didn’t spend time second-guessing earlier decisions. “I never sat there and said, ‘Oh, crap, why do we have so much commercial real estate?’ ” he says.
Instead, he started thinking about data. Many of GE’s corporate customers were putting sensors on their machines to collect information about them. That often meant a lot of information: A jet engine, for instance, spits out roughly a terabyte’s worth of everything from fuel usage to heat levels to the size of the specks of dirt that fly through the engine on a trip across America. What were GE’s customers supposed to do with all that data? Immelt considered teaming up with a tech company to create software that would analyze vast amounts of the stuff, but when the tech company figured that out, what would it need GE for? He thought GE would be better off developing this software on its own. If nothing else, the company would be able to use the technology to improve its own productivity; if things went well, GE would be able to sell it as an add-on to service contracts with industrial customers. “I said, ‘Look, we need to start building analytic capability, big data capability, and let’s do it in California,’ ” Immelt says. “That was as sophisticated as my original thinking was.”
His own knowledge of the software business was limited. Along with doing whatever it took to win a character award at Dartmouth, he graduated with a dual degree in economics and applied math in 1978. After getting his MBA at Harvard, he turned down a job at Morgan Stanley to work at GE. He ended up running the health-care division in the 1990s, which opened a software center of its own in Wisconsin. Because of that, he says, he knew enough to ask the right questions about software.
If GE was going to set up shop in Silicon Valley, Immelt wanted a local to run the operation. He went after William Ruh, then a vice president at Cisco Systems. Ruh was astonished to get a call from a recruiter who was coy with him about the company she represented, asking him to guess which one it might be. “I named a name, and she said no,” he says. “And I named another name and she said no. She said, ‘Name another.’ I said, ‘No, I can’t name anymore. Just tell me, who is it?’ And she said, ‘GE.’ I said, ‘It can’t be. They don’t know anything about software.’ ”
Ruh found it hard to believe that GE would be willing to invest the kind of money it would take to build a successful software business. Silicon Valley is full of little startups, but creating software at an industrial scale would require billions of dollars. He also couldn’t see GE, with more than 300,000 employees, making the cultural changes needed to compete in the Valley.
Despite those qualms, he traveled to Fairfield in January 2011 to meet with Immelt. Ruh says he was impressed by Immelt’s vision and his willingness to admit that he didn’t fully know what he was doing. “Basically, Jeff said, ‘Look, we’re on Step 1 of a 50-step process, and I just need you to help me figure out what to do because I can only see out one or two steps,’ ” Ruh says. He took the job, and several weeks later his new boss promised to invest $1 billion in a software operation in San Ramon, Calif.
GE’s ambitions were greeted with skepticism in the Valley. In 2012, when Immelt promoted the software venture in San Francisco during a company-sponsored event with Marc Andreessen, the star venture capitalist and a friend of Immelt’s warned that it would be difficult for a hardware company like GE to assemble a team of data scientists that could perform the kind of tasks that GE had in mind. “It’s hard to be really good at that,” Andreessen said. “It’s really complicated.” (Bloomberg LP, which owns Bloomberg Businessweek, is an investor in Andreessen Horowitz.)
Jennifer Waldo, GE’s head of human relations in the San Ramon office, says recruiters had a hard time just getting people to come in for an interview. Nine out of 10 software developers they contacted had no idea GE was in the business or that it even had operations in California. Nor were they necessarily interested in learning anything more: Almost all had jobs and couldn’t see any upside to working for an East Coast microwave oven manufacturer. In 2013, Waldo appealed to Immelt for help when he visited San Ramon. “I walked him through all those issues,” she says. “I needed to compensate differently. I needed to in-source my recruiting team. We were competing in a marketplace where we’re not even a recognized player.” A former GE recruiter says the company offered stock options to job candidates, but not actual stock, the norm in Silicon Valley. There were also no nap rooms, no on-site child care, no dogs wandering around the office.
Waldo and her team found they could make headway by telling prospects that they would have a chance to develop trains and power equipment rather than some inconsequential social-networking app. “I had a candidate in the early days,” she recalls. “She came in and said, ‘I’m sitting there trying to figure out how to put a Pinterest button on something, and I get this phone call from GE, and you’re talking about making aircraft engines fly more efficiently.’ ” (A GE spokeswoman says the company now includes stock in its compensation for software developers, too.)
GE also targeted startup veterans who’d spent years putting in hours for low pay hoping to be the next Mark Zuckerberg. “They went around to guys who were in their second and third startup and had been eating ramen noodles for eight years,” says Nick Heymann, an analyst at William Blair. “They said, ‘Look, how would you just like to have a normal lifestyle, live an hour outside the Bay Area, make a quarter of a million bucks a year, and give your kids a really good education?’ ” At the end of 2013, GE had 750 people working in San Ramon.
By then, GE had developed an early version of Predix, an operating system like Windows or Android but for the Industrial Internet. The company developed applications for Predix enabling it to ingest and analyze vast amounts of data from sensor-equipped machines much like Amazon.com, Facebook, and Google do with information generated by their human customers. Immelt wanted to speed Predix’s development and use it on GE’s own equipment. That meant the entire company had to embrace the new operating system, even the power division, which usually took years to design turbines. There didn’t seem to be much need to rush out new models; GE’s power customers typically buy steam- or gas-powered turbines and use them for three decades.
The more Immelt watched what was happening in Silicon Valley, the more he became convinced GE needed a cultural revolution. He sought assistance from Eric Ries, a tech entrepreneur and author of The Lean Startup, a book that espouses the importance of releasing early versions of products, getting customer feedback, then “pivoting” or changing them if necessary to improve them. In 2012, GE asked Ries to speak to Immelt and some of his top executives at the Ossining training center.
Ries was so nervous that he wore a suit. When he arrived at the training center, he says he felt like he was entering an alternative universe. The day before, he’d been in Washington visiting members of the Obama administration. Yet when he mentioned the White House with the GE people, they thought he was talking about the building on campus where the bar used to be. “I’m a startup guy from San Francisco,” Ries says. “I was just like, ‘What on earth is happening here?’ ” Ries was expecting Immelt to be a brusque, Jack Welch-like character. Then the CEO showed up in jeans and kidded him about being overdressed. “ ‘I thought you were from Silicon Valley,’ ” Immelt told him. “ ‘What are you doing in a suit?’ ” Ries was charmed.
After Ries gave his presentation to the group, he opened the floor to questions. There was an awkward silence. “Jeff turns around, and he names one of his vice presidents, and he says, ‘How come you’re not already doing this?’ ” Ries remembers. “The guy was like, ‘Um, mumble-mumble-mumble.’ All of a sudden, there were a lot of questions in the room. It was like, ‘Message received. Jeff thinks there’s something here.’ ”
That afternoon, Ries started giving workshops for executives. He later helped GE tailor its own version of his methods, which the company calls FastWorks. He says Immelt wanted change, telling him: “ ‘I’m tired of hearing five-year plans.’ ” GE has since handed out thousands of copies of The Lean Startup and has trained tens of thousands of employees in the process. Everyone in upper management seems to use Silicon Valley-compliant vocabulary, particularly the word “pivot.” “We encourage people to try things, pivot, try them again,” Immelt says. “It’s a better way to run the place than centralized command and control, process-laden.” It’s a sign of his personal charisma that, unlike many of his employees, the boss can speak corporate jargon and make it sound profound.
Immelt also thought it was time to revamp GE’s annual review process to make the company more palatable to younger, software-literate workers. Under Welch, GE was famous for annual reviews that ranked all its employees numerically according to their performance. Then the bottom 10 percent were fired. “The biggest cowards are managers who don’t let people know where they stand,” Welch told Bloomberg Businessweek in 2012.
“It just didn’t make sense anymore,” says Susan Peters, GE’s senior vice president for human resources, of the annual review process. The company decided to scrap them altogether, replacing them with a gentler system where employees are “coached” by their more experienced peers.
Unlike some of Immelt’s earlier management initiatives—the idea jams and the imagination breakthroughs—the new ones seemed to have the intended effects. FastWorks, according to GE, enabled the development of a new gas turbine in a year and a half, rather than the usual five. “This is a billion-dollar product line,” says Steve Bolze, chief of GE’s power division. “It’s going to expand to be one of our single biggest launches in our history.”
In April 2015, Immelt announced his plan to sell $200 billion of its GE Capital assets within two years, faster than Wall Street had expected. Not surprisingly, he called it a “pivotal day.” Previously skeptical analysts praised Immelt during a conference call when he explained the plan. Even Barclays Capital’s Scott Davis, who had speculated only a month earlier that Immelt would soon be out of a job, was contrite. “Congrats,” Davis said. “I know we have all given you a lot of crap over the years, but this is pretty good stuff for redemption. That’s my best apology.” He added, “You can keep your job a little longer, I guess.”
Along with unloading most of its risky financial business, GE struck a deal to sell off its appliance division to Haier, the Chinese conglomerate, and increased its share of the global power industry with its 2015 purchase of Alstom, a French energy company, for $10 billion. As a result, 90 percent of GE’s profits will come from industrial operations by 2018. (During its glory years from the late ’90s into the mid-2000s, GE Capital contributed as much as 40 percent.) Immelt says this is the GE he’s been trying to create since the financial crisis, although he acknowledges that it might have been difficult for outsiders to discern.
In January, GE announced the move to Boston, where the deer are few and the software developers plentiful. “Sitting in a rural setting, you can never be scared enough of what’s next,” Immelt says. “You just can’t be. You can’t be paranoid enough. And I felt like it would be a good thing for the business just to be in the flow of ideas.” It could also help GE attract more young employees with technology backgrounds, which remains a struggle because most people still don’t associate GE with software. GE has acknowledged this by running TV ads featuring a fictitious coder named Owen who gets blank stares from his friends when he tells them he’s been hired by GE.
“Guys, I’ll be writing a new language for machines so planes, trains, and even hospitals can work better,” Owen says.
“So you’re going to work on a train?” a friend asks.
“No, on trains,” Owen corrects him. “On trains.”
“So you’re not going to develop stuff anymore?”
It’s rare for GE to laugh at itself like this, but the commercials have helped recruiting. “The Owen ads have increased the number of résumés we get by eight times,” says Ruh, now GE’s chief digital officer.
Head count at the San Ramon office is 1,300, including some refugees from Google and Facebook. It already has aviation customers using Predix applications to monitor the wear and tear on their jet engines and calibrate their maintenance schedules based on that data rather than an average for the entire fleet. It’s created smart wind turbines that tell each other how to shift their blades to catch more wind, which GE says can increase their power output by as much as 20 percent.
But Immelt needs to sell vast amounts of applications and Predix-based analytics to reach his goal of making GE a top 10 software company in 2020. That’s not a random deadline. Traditionally, GE chief executives have served 20 years, so his time will be just about up by then. He says the company already has a succession plan in place. If he can complete his digital reinvention of the company, he could depart in glory, the way Welch did.
GE says it’s beginning to sell Predix-based services to customers who design their own industrial equipment. Pitney Bowes is using Predix on its mailing-label machines and letter-sorting devices in corporate mailrooms; Toshiba is using it on elevators. “The Industrial Internet is going to be the dark matter of the Internet,” promises Harel Kodesh, chief technology officer for GE Digital, which is what the company now calls its software division. “It’s something you don’t see, but it is actually the bulk of what’s happening on the Internet. Other than porn, I guess.”
That may be, but GE faces competition from all sides. Amazon and Google are getting into the Internet of Things along with IBM and Microsoft. There are dozens of small startups with similar ambitions that don’t need Eric Ries to tell them what to do. Then there’s perhaps an even bigger unknown: Will other large industrial companies turn to GE to manage their information? “If you’re a manufacturer of some size and sophistication, are you going to say, ‘Hello, GE. You can own the data on my business’? ” asks Brian Langenberg, an independent analyst and founder of Langenberg & Co. in Chicago. “Or are you going to say, ‘I think I’m going to do it on my own’? I’m skeptical.”
Not long ago, Immelt gave a speech in Dubai about the Industrial Internet. He wasn’t talking to tech people “with spiked hair,” as he puts it. He was addressing attendees who worked at oil companies and airlines—in other words, his kind of people. Immelt could see them nodding their heads approvingly as he talked. “It’s moments like that when you think, ‘This might work,’ ” he says later. “ ‘This really might work.’ ”