The first analyst to offer a comment was Scott Davis of Barclays, who for years had been slamming Immelt and his sorry record of wealth creation and who at one point had called for the CEO’s ouster. But now his tone was transformed. “Congrats,” Davis said on the conference call. “I know we have all given you a lot of crap over the years, but this is pretty good stuff for redemption. So thanks for that. That’s my best apology I can make, by the way.” And then, to general laughter: “You can keep your job a little longer, I guess.”

When trading opened soon thereafter, GE’s (GE) share price—which till then had slid a dismal 36% during Immelt’s tenure—shot skyward, rising 11% by the close. The CEO had increased GE’s market value by $35 billion in one day with a press release.

GE’s move is the most dramatic example yet of a highly significant, six-syllable trend: deconglomeration. The world’s biggest diversified conglomerates are finally realizing that combining entirely dissimilar businesses in one company almost never works. For the first time in living memory, GE is on the road to becoming a coherent whole, built around industrial infrastructure businesses including power generation, locomotives, jet engines, and oil and gas production equipment. As chronically skeptical investors understood immediately, that should be good news.

GE Capital was founded in the 1930s to help consumers buy GE refrigerators and other appliances but had blossomed into a mammoth profit machine with almost no relation to the company’s other businesses. It was America’s seventh-largest bank. It owned office buildings and stores; financed supermarkets, fast-food franchises, and other mid-market businesses; loaned money to consumers; sold insurance; and at one time even made subprime mortgages. Disposing of its $200 billion of assets will take until sometime next year. Only the operations that help industrial customers buy GE products will remain.

As another step in deconglomerating, GE plans to sell its appliance business to Electrolux, though the U.S. Justice Department has sued to stop the deal; GE says it intends to “vigorously defend” it in court. And emphasizing its new focus with its biggest acquisition ever, the company agreed to buy the electric-power assets of France’s Alstom for some $13 billion. The European Commission has expressed concerns, to which GE has proposed remedies. The commission promises a decision by Sept. 11.

Why do investors cheer when a company off-loads a huge, profitable operation? Because if that business is unrelated to the company’s other businesses, it will probably be worth more to some other company that’s in its business, or on its own. A seemingly sudden realization of that fact—plus a friendly environment for selling or spinning off assets—has sparked an epidemic of breakups: News Corp. separating its print properties from its entertainment businesses, eBay (EBAY) spinning off PayPal (PYPL), Hewlett-Packard (HPQ) separating its PC and printer business from its corporate hardware business, United Technologies (UTX) selling its Sikorsky helicopter unit to Lockheed Martin (LMT), and dozens more such moves.

The truth is that diversified conglomerates were always a bad idea. As a concept they were hot stuff in the 1960s, when investors stampeded to own ITT (ITT), Litton Industries, Gulf & Western, and LTV. All those companies are gone because they were worth more in pieces than as a whole. Yet conglomerates persisted, despite the evidence. McKinsey research finds that from 2002 to 2010, conglomerates’ median total annual return to shareholders was 7.5%, vs. 11.8% for more focused companies.

The world mostly shrugged at the breakups of Gulf & Western and Litton. But GE has always been different. It’s an American institution, founded by Thomas Edison and revered for virtually its entire existence. Even today its leadership-development programs earn it the No. 1 spot among Towers Watson’s Top Companies for Leaders. So it’s disorienting to realize that even GE has been, overall, a typical crummy conglomerate.

Its conglomerate period began in 1945, after World War II, when the company expanded its plastics business and became, over time, a major player in jet engines, nuclear reactors, computers, missiles, satellites, and other widely diverse businesses. From then until now, GE’s total shareholder return has slightly beaten the S&P 500’s—not so bad a record, it seems. But look closer. All of that slender outperformance and much more is attributable to one CEO, Jack Welch. Remove the Welch effect by assuming GE had matched the S&P during his tenure, and the stock would have massively underperformed the S&P over its 70-year conglomerate period. Except for a once-a-century managerial phenomenon, GE has performed no better than many other conglomerates that were broken up long ago.

You can see why investors were delighted that GE is finally abandoning the strategy of sprawl. The puzzle they face now is how to evaluate what Immelt has called “the new GE.” How much do the remaining parts really have in common, and what are their prospects as a combined entity?

GE is now far more focused than it has been in many decades. The most obvious commonality of its remaining units: spinning. Jet engines, power turbines, motors, drives, the inside of a locomotive, a CT scanner—they all spin. The technology overlaps are valuable: For example, carbon-fiber-reinforced composites developed for jet-engine fan blades are used in GE’s wind-turbine fan blades, and ceramics developed for electricity-generating turbines are used in jet engines.

A more important commonality is not so obvious: Virtually everything GE will do after its reordering contributes to an economy’s basic physical infrastructure. The company’s strategists observed that the world’s advanced economies are scarcely growing, while the developing and even some non-developed ones are growing fast; though China and others are slowing, together they still create more new GDP every year than do the more established economies. And those nascent economies need infrastructure—electricity above all, transport across land (rail is most efficient) and in the air, oil and gas, public lighting, and basic health care.

Often all those goods and services are bought by one customer, a government, or companies that are officially or unofficially government-controlled. GE wants to be, among other things, the one-stop shop for nations entering the developed world. “We’ve been able to speed our efforts in individual markets,” says Beth Comstock, who runs GE’s business-innovations unit. “Every business doesn’t individually have to figure out, ‘How do we all go to Ethiopia?’ for example. An advance team goes out and understands how we pull all the pieces together.”

Do those factors make for a coherent business? Increasingly they do. Consider, for example, a factory that the company opened this year in Pune, India. It makes components for four different GE operations—jet engines, locomotives, wind turbines, and water-treatment units—the first time a factory has ever done that. Some of the plant’s equipment, such as 3-D printers and laser inspection technology, can be shared. Making components for four businesses in one plant is more efficient than building freestanding facilities for each one. It’s also the first test of what GE calls the “brilliant factory” concept, with the plant’s equipment, suppliers’ machines, and distributors’ machines all communicating continuously on what GE calls the industrial Internet. “You’ll see that more and more,” says Comstock. And the plant contributes significantly to the “Make in India” campaign in the world’s second-largest developing economy. Prime Minister Narendra Modi attended the opening.

GE has always tried hard to transfer useful ideas among its many units. The question is whether the benefits outweigh the drawbacks of housing diverse businesses under one roof. With the company’s portfolio now more focused, the odds improve. GE calls the sharing concept the GE Store, and it’s one of the company’s main arguments for why its parts are more valuable together than separated. “Let’s say there’s a subsea development of a field,” says Lorenzo Simonelli, who runs the oil and gas business. “We’re working with the GE Store on the large motors that have to come from energy management, with power and water on the gas turbine, with aviation because we’re going to need aeroderivatives [gas turbines based on jet engine technology]. We’re working through the Store with health care because there’s going to be pipeline inspection”—using X-ray technology—“and potentially with financing, which is the vertical that we have through GE Capital that remains. On large-scale projects, the GE Store really comes to life.”

Although GE initially becomes smaller by jettisoning most of GE Capital, its new strategy is grand in scale. “Until now, no company has ever attempted to become a one-stop source for helping modernize undeveloped and underdeveloped regions of the world,” says Nicholas Heymann, an analyst at William Blair who has covered GE for some 30 years. “To attempt this was simply unfathomable for even the most ambitious and well-resourced companies in the world.” GE is telling investors they’ll have to wait two years to start seeing a financial payoff. That’s probably why the stock, after spiking on the announcement, has drifted back down. Investors are prudently waiting for evidence of success.

The corporate transformation is so pervasive that GE is even altering its famous culture. “We could be doing all of these incredibly important and strategic portfolio shifts, and if the culture doesn’t simultaneously shift to reflect what’s happening in the world, then we may be at odds,” says HR chief Susan Peters. They call the new ideal a culture of simplification, based largely on the 2011 book The Lean Startup by Silicon Valley entrepreneur Eric Ries (with a new foreword by Jeff Immelt). The goals are speed, empowerment, and minimal bureaucracy, a tall order in a 137-year-old company with 305,000 employees, but necessary.

Let’s assume the corporate refocusing comes off as planned. GE will still be a fairly diverse company. Part of that is unavoidable. For example, the company is becoming a major player in the software business—seemingly a very distant cousin of those heavy industrial businesses—because it must. Google (GOOG), IBM (IBM), and other giants are offering to analyze the massive data streams pouring out of industrial equipment, potentially cutting GE out of a high-value service for its customers. So GE is playing offense with a 14,000-employee software operation and a newly announced service that performs predictive analytics on anyone’s equipment, not just GE’s. Integrated with the other businesses, software (forecast 2015 revenue: $6 billion) is looking like an excellent performer. It earns 50% operating margins, requires little capital, and is growing fast.

As dramatic as GE’s new focus is, you can’t help asking whether it goes far enough. PET scanners and undersea oil-well blowout preventers are still pretty different, for example. Could more changes lie ahead? GE Lighting had a for sale sign on it before the financial crisis, and Credit Suisse analyst Julian Mitchell calls it a “logical candidate” for divestiture. But as LEDs become connected devices, GE says the business makes sense in an industrial infrastructure portfolio. Barclays’ Davis has observed that “some investors want a spin-off of medical,” but GE has not breathed a syllable to encourage them.

Bottom line, will the deconglomerated GE work? The strategy sounds plausible and encouraging, but GE strategies have always sounded that way. What we can say for sure is that GE is transforming itself more fundamentally than it has done in at least 30 years, and the movement is all in the right direction. Shareholders should hope the company achieves its newly stated goals swiftly and adeptly—and then goes even further.

A version of this article appears in the September 1, 2015 issue of Fortune magazine with the headline “Breaking up is hard to do.”