If you google the word conglomerate, you’ll learn that its original meaning was geological: basically, a rock that’s composed of different bits of rock, held together by some sort of clay, silt, or sand. Its corporate sense—“a hodgepodge of different enterprises all roped together under one name,” as Life magazine once explained it—burst onto the scene in the 1960s, when, for a brief and madcap moment, sticking Paramount Pictures, Consolidated Cigars, Beautyrest mattresses, and the New York Knicks into an amalgam called Gulf + Western seemed to make sense. It didn’t, and by the 1980s, this second meaning had become a corporate slur. “If you mix beef broth, lemon juice, and flour, you don’t get magic,” the conglomerator Harold Geneen wrote in 1997 after his Twinkies-and-rental-car empire, ITT, had returned to nature. “You get a mess.”

Google will also inform you, of course, that Google has decided to reorganize itself into “a General Electric–like conglomerate called Alphabet.” Those were The New York Times’ first words on the announcement, and they’re loaded not only with that dirty word, conglomerate, but with two very sparkly ones: General Electric, the name of the most consistent corporate performer of the 20th century and the rare conglomerate that both predated and outlived the craze for them in the ’60s and ’70s.

What to make of Google’s future? To answer that question, it helps to first understand why GE isn’t out in the conglomerate boneyard.

What put the other conglomerates there is no great mystery. Gathering dissimilar businesses under a common roof was supposed to make each more valuable. They could balance out one another’s risks, borrow cheaply with a common credit card, and share access to the superior management minds at headquarters. But it actually made them less valuable. Avis Rent a Car didn’t have anything to teach Continental Baking, and vice versa, so neither grew any faster. The management minds at headquarters couldn’t allocate capital as well as the market could. And by the late 1970s, investors looking for diversification could simply buy a mutual fund. Shareholders tired of the poor performance. ITT began unwinding itself in the mid-’80s. The lavish Gulf + Western building was stripped to its skeleton and renamed Trump International Hotel & Tower about a decade later.

GE’s persistence is harder to explain. The company is in the midst of unloading most of GE Capital (a major drag on its stock price since the 2008 financial crisis) and its home-appliance business. But even after that’s done, its earnings will come from lightbulbs, locomotives, software, wind turbines, jet-engine maintenance, ultrasound equipment, and all the other products of a “multi-business industrial company”—the phrase GE executives gently suggest in place of conglomerate. This portfolio produces a river of profits that in the eyes of most analysts makes GE worth more than the sum of its parts.

So what is GE’s secret sauce? The most common answer is the one I got from Jeffrey Immelt back in 2000, a few days after he’d been named Jack Welch’s successor as CEO. We were sitting in GE’s famed management-training facility in Crotonville, New York. What, I asked, was GE’s core expertise? Immelt responded, “We pick great people, we make them better, and we retain them.” Only after a second or two of silence did I realize that this was his whole response. But really, no elaboration was necessary. As we met, Immelt’s two defeated rivals for the top job were being scooped up to run Home Depot and 3M. GE’s unmatched record for cultivating management talent spoke for itself.

The “talent factory” theory of GE bodes well for Google. It, too, is celebrated for a rigorous people-development process that has produced, among others, Facebook COO Sheryl Sandberg.

But three years after our first meeting, I met with Immelt again—and heard him articulate a second, lesser-known ingredient in the secret sauce. Actually, I coaxed him out onto the roof of Rockefeller Center, to stand just above the giant GE letters. What he told me out there, I’m not sure. (“God, this is something” is all I could make out above the wind.) Back inside, though, Immelt talked about “big bets,” and the importance of making the kind of bets that only GE had the wherewithal to make. He was invoking an earlier epoch of GE history, mostly forgotten, when the company was less about scientific management and more about science.

This is where the Google comparison gets interesting. Inside the House of Magic—the nickname for GE’s research lab in Schenectady, New York, which opened in 1900—some of the most eminent scientific minds of the day were set loose on far-out projects. Some, like an electric-car charger, proved as ill-fated as Google X’s explorations of hoverboards and teleportation. But others, like the X-ray tube, opened up entirely new fields: radio, medical imaging. There was no one place on the balance sheet where the House of Magic’s impact could be seen, GE’s chairman, Owen Young, said in 1930. “It is the balance sheet itself.”

The house of magic never disappeared, explains Christopher Hunter, the director of archives at Schenectady’s Museum of Innovation and Science. Rather, it disappeared from view. In the 1950s, the management-training facility in Crotonville took its place as GE’s most heralded incubator. But the managers created at Crotonville, including Welch and Immelt, were the inheritors of a corporate success story built from the company’s forgotten product lab—which Immelt has revived and reemphasized.

For Google—launcher of Earth-imaging satellites, driverless cars, and a crusade to end death—the “innovation hothouse” theory of GE bodes well, too. General Electric was clearly the Google of its day. Is it premature to call Google the GE of the 21st century?

Actually, yes. GE makes money lots of different ways. Google does many things, but it makes almost all its money one way: through online advertising. This business accounts for nearly 90 percent of the company’s revenues. Everything else—YouTube, Android, Chrome, Nest, Fiber, Gmail, Google Earth, Google+, Google Drive, Google Docs, Google Maps, Google Capital—contributes just 10 percent. Google, financially speaking, is a monolith.

Bear in mind that, as late as the 1930s, lighting accounted for up to two-thirds of GE’s profits. But by the time the lighting business lost its pop as a profit engine—as maturing businesses are wont to do—some of the company’s other projects had kicked in as profit engines in their own right. This points to a third, vital ingredient in the GE recipe that Google has yet to obtain: mastery of the very difficult process of turning cool science projects into commercially viable products. Without this capacity, GE would have burned out long ago. And without this capacity, Google will eventually risk comparison to another 20th-century icon—Xerox.

Like Google today, Xerox in the 1970s was a thrilling tech company whose massive profits gushed from a narrow source, the photocopier. Intent on extending its technological lead, Xerox poured some of those profits into Xerox PARC, the lab it created to develop the technologies of the future. But Xerox’s copier business lost steam before any of the revolutionary technologies that PARC pioneered—the mouse, the Ethernet, the overlapping windows and icons you see on your computer today—were successfully commercialized. (Apple, famously, was the company that eventually did much of that commercializing, after Xerox blithely invited Steve Jobs in for a tour.)

What was the difference between GE and Xerox in this regard?

The inner workings of each company’s R&D process, fortunately, have been the subject of detailed histories. Dealers of Lightning, Michael Hiltzik’s history of Xerox PARC, contains numerous cautionary tidbits for Google. For one, relying on moon shots while belittling incremental innovations is dangerous. Second, it’s not true that “if you have the smartest people, you’ll end up ahead,” as one PARC executive flatly professed. What comes across in Hiltzik’s book, above all, is the mutual sense of alienation within the company. Xerox’s businesspeople didn’t speak the language of science. Its scientists didn’t speak the language of business. And neither side seemed interested in learning. Complaining about the idiots in corporate or the flakes at PARC was easier. “So few of us accepted responsibility” for overcoming that language barrier, observed Bob Metcalfe, a co-inventor of the Ethernet at Xerox PARC, who went on to found 3Com.

Both history and geography may have contributed to the problem. PARC was established with money from a product monopoly it played no role in creating. And it was located in Palo Alto, California; Xerox, in Stamford, Connecticut.

GE’s Schenectady researchers, by contrast, were wedged in cheek by jowl with manufacturing, sales, and every other corporate function. And job No. 1 in the early days at the House of Magic was securing GE’s then-tenuous footing in the lighting business. It delivered big in 1910, with the tungsten-filament lamp—an innovation that conveyed decisive control of the lighting market to the mother ship.

The steady profits this ensured allowed the lab to widen its scope of exploration. But being thrust, very early, onto the front lines of the company’s main commercial battle also left a lasting cultural effect. The effort forced scientists to think like industrialists and industrialists to think like scientists. In the decades that followed, manufacturing people would routinely ask for lab input on this or that problem. And the researchers, to test the viability of this or that contrivance, weren’t averse to building small, working factories within the lab.

“The company is not primarily a philanthropic asylum for indigent chemists, and I must not let it become one even secondarily,” wrote the lab’s guiding hand, Willis Whitney—who, at the same time, was known for popping his head into offices to ask “Are you having fun?” There was, to be sure, plenty of friction between the commercial imperative on the one hand, and the joy of discovery on the other. But the magic lay in their uneasy coexistence—something GE understood and embraced.

The most worrying thing for Google, then, should be the lack of commercial imperative in so many of its operations. In that light, the most promising thing about Google is its reorganization into Alphabet—a step that will inevitably put pressure on its newly autonomous divisions to produce something besides giant losses. Nest (Google’s push into the automated home) and Fiber (its push to provide super-fast cable and Internet services) seem closest to becoming what Google’s executive chairman, Eric Schmidt, has called “real businesses.”

Steve Jobs once warned Larry Page, a Google co-founder, that Google was in danger of becoming Microsoft—a company caught, maybe perpetually, in the Xerox trap. Perhaps Alphabet is Google’s way of putting a gun to its own head and forcing itself to become something more than an interesting collection of rocks.