(FORTUNE Magazine) – You may not be able to see it, but you can't avoid its buzz. Nanotechnology is fast becoming as pervasive a cultural icon as TiVo or Levitra. The wizardry of building teeny things that are measured in one-billionths of a meter has begun to figure in Hollywood movies, in bestselling novels--even in Jay Leno's monologues. Nanotech has also inspired a dithyramb on the website of the ordinarily sober National Science Foundation: "Imagine a medical device that travels through the human body to seek out and destroy small clusters of cancerous cells before they can spread. Or a box no larger than a sugar cube that contains the entire contents of the Library of Congress. Or materials much lighter than steel that possess ten times as much strength." No wonder nanotech generated a $3.7 billion federal funding commitment in December.

Nanotech has electrified the investing world too--so much so that in the past six months cautionary articles have appeared in the Washington Post, the Wall Street Journal, USA Today, the Baltimore Sun, Investor's Business Daily, the Boston Globe, and other publications. Each has called nanotech an emerging stock-market bubble and warned readers not to be swayed by the hype. You'd think investors would hesitate to give wild valuations to startup technology companies whose revenues are even more minuscule than their products. But many of us, it has become clear, have shorter memories--and stronger conditioned responses--than a lab rat.

Of course, Wall Street professionals aren't exactly models of prudence and restraint either. Eager to catch the wave, Merrill Lynch announced--on April 1, fittingly--the creation of a nanotech stock index. Projected backwards, it showed that the stocks had risen 120% in the previous 12 months, compared with 31% for the S&P 500. Only 15 days later, Merrill overhauled the index. It turned out that nearly a quarter of the companies on the list weren't really in the business of nanotechnology--or at least so far as the companies themselves were willing to say in their public filings.

And there, it seems, is the rub. While nanotech is unmistakably big, nobody seems to agree on what it is. That includes the practitioners and financiers of the tiny science, who vehemently disagree on its definition, or even its usefulness as a business category. Consider the view of venture capitalist Steve Jurvetson, of Draper Fisher Jurvetson in Menlo Park, Calif. A thin, enthusiastic 37-year-old, Jurvetson has pumped $88 million of clients' money into 20 nano and microtech startups, and is sometimes referred to as Mr. Nanotech. He describes nanotech as the next industrial revolution and predicts it will permeate every form of manufacturing. Yet he scoffs at the idea that the term is a meaningful way to classify a business. "Nanotechnology as a descriptive phrase is as absurd as 'incho-technology'--all the companies that build things on a one-inch scale get lumped in a bucket, right? It's just a laughable concept."

Yes, nanotechnology is a realm rich in innovation--and in paradoxes. It is simultaneously a bubble and a puzzle. It is at once brand new and age old, with sales in the tens of billions of dollars or with no sales at all--depending on whom you ask. It inspires hype over its dazzling future, yet the hype has obscured its surprising--though altogether more prosaic--accomplishments in the present.

And here is perhaps the most striking paradox: Though nanotech is driving a wave of startups, many of its achievements are emanating from staid giants, the very companies that surges of innovation are supposed to leave behind. If you think nano is the preserve of nerdtrepreneurs, listen to Stan Williams, director of quantum science research at Hewlett-Packard: "The world's premier nanotech company is General Motors." Clearly, whatever it is, nano deserves a closer look.

Nobody disputes the fact that a nanometer--to which the "nano" in nanotech refers--is one-billionth of a meter, or about 1/100,000 the width of a human hair. Scientists and engineers working at this scale have the ability to manipulate individual atoms and molecules. But beyond that, there are any number of diverging definitions of nanotechnology. Two are most common. What could be called the broad vision embraces any material or device that has features of 100 nanometers or less. The narrower definition is trickier: In this view something qualifies as nanotech only when manipulation at the nanometer level changes the physical properties of whatever is being manipulated in ways different from what would occur at a larger scale. It's where, as some scientists put it, Newtonian physics gives way to quantum physics. So under the stricter definition, shrinking a semiconductor part to 90 nanometers (as Intel now does) doesn't count. But using a laser to trap and position individual molecules (as a Chicago startup named Arryx does) would qualify. Arryx makes equipment to manipulate micro and nano particles.

Among scientists in the field, the standard joke has become, Whatever I'm doing is nanotech, and whatever he's doing isn't. By the broad definition there are tens of billions, if not hundreds of billions, of dollars in nanoproducts on the market. Last year Intel alone sold $20 billion of Pentium 4 and other chips, all of which have circuit elements smaller than 100 nm. But if you use a more restrictive definition, well, that figure could be closer to $1 billion or $2 billion, or even zero for some hard-liners. "No one today is in a position to do nanotechnology," declares Eric Drexler, the theorist whose 1986 book, Engines of Creation: The Coming Era of Nanotechnology, thrilled geeky readers with visions of self-assembling nanomachines and helped launch the field. "When I introduced the term," Drexler says, "I intended it to be applied to a capability that we wouldn't get to for decades, and indeed, we're not there yet."

Still, however you define it, nanotech is becoming harder to view solely as something that won't turn up for decades. It's here now.

Congratulations: the miracle of the future has arrived and it's ... stain-resistant pants? That's right. Today Levi's Dockers and Gap khakis employ nano-fibers, which bind to the cotton fibers in the fabric to create an invisible barrier that protects your slacks from the depredations of spilled red wine or coffee.

There's more. Chances are your computer's hard drive has an IBM-designed nanocomponent known as GMR, which has dramatically improved memory performance. (GMR technology hinges on a nano phenomenon called the giant magnetoresistive effect; we'll spare you the details.) Running boards in some General Motors vans and SUVs have been bolstered with nanofillers for years; as of January, Chevrolet replaced the "rub strip"--the molding that protects car doors from other car doors--on the Impala with a nanocomposite version. L'Oreal blends liquid nanoparticles into face creams to help moisturizer penetrate deep into skin. And Eastman Kodak has begun marketing a digital camera whose viewscreen includes a nanocomponent.

So that's what all the buzz is about? Instead of a cure for cancer, we get protection against being dinged in a parking lot? It seems absurd. Yet none of this is actually bad news. It's true that with few exceptions, nanotechnology has advanced only far enough to develop materials rather than more complex devices. And nanotechnology at present tends to consist of one nanomaterial or component built into a non-nano product. Says Sean Murdock of AtomWorks, a nano business advocacy group in Chicago: "Companies are doing the BASF model: We don't make things. We make things better."

Incrementalism is a helpful lens through which to view the nanotech trend. It's an indication that we are, in fact, on a road that may--may--one day lead to the wondrous breakthroughs that are the stuff of National Science Foundation fantasies. But we might never get to cancer-curing nanobots if we hadn't developed stain-free trousers first.

To see forward, it helps to look back. You could argue that today's nanotech is a step in a long evolution. "Advances in information technology have been going on for 100 and some years," says Tom Theis, director of physical sciences at IBM Research, "and it's all been about miniaturization." Like many of the scientists interviewed for this story, Theis seems more concerned with tamping expectations than with touting his work. "We were publishing papers at IBM on nano-structured materials in the 1970s," he says. "None of us in this field feel that there's anything fundamentally new here. It's the logical culmination of the long-term historical trends in miniaturization of many technologies."

That doesn't mean there's not remarkable new value. It's just that nanotech didn't alight unexpectedly from another galaxy last week. Says Howard Berke, the chairman and CEO of Konarka, a nano startup in Lowell, Mass.: "It's really fulfilling the vision of folks like Albert Einstein and Thomas Edison," who long ago anticipated how the laws of nature might ultimately affect technology. The scientific principles haven't changed--only our ability to put them into practice has. Konarka itself is a perfect example. The company is applying nanotech to photovoltaics--the conversion of sunlight into electricity. Photovoltaics is more than a century old; Berke is quick to point out that Einstein won his Nobel prize not for his theory of relativity but for his work explaining photovoltaics. But only now is it becoming feasible to, say, fabricate otherwise ordinary roofing shingles that deliver electric power--one of Konarka's goals. The company makes nanocrystals coated with light-absorbing dye and embedded in an electrolyte; they can be layered onto plastics, foils, textiles, and other surfaces. Konarka plans to have its first product, whose applications include providing solar power for security sensors, on the market next year.

For chipmakers like IBM's Theis and Paolo Gargini, director of technology strategy at Intel, nanotech is simply the next step in the progression of Moore's law. They see it not as an option for their industry but as a requirement. Chips are getting close to the point at which they no longer can be shrunk--not because circuits can't be manufactured at a smaller size (though cost does become increasingly prohibitive), but because the structures will stop behaving the way they do at larger scale. Electron leakage between adjacent circuits can pose problems, for example.

In experiments, Intel has harnessed nano magic to overhaul an insulating layer in its Pentiums and other chips. By the weird rules of quantum effects, adding three molecules to a one-molecule layer makes it thicker--it prevents current from leaking--without actually taking up more space. Other chipmaking giants like IBM, Hewlett-Packard, and Motorola are working feverishly on their own solutions to similar problems.

Startups are also racing to marry chipmaking and nanotech. With backing from venture-capital powerhouse Kleiner Perkins Caufield & Byers, Zettacore in Denver is developing prototypes of so-called molecular memory chips. Instead of storing charges in silicon, as traditional memory chips do, Zettacore plans to store them in nanostructures composed of porphyrin, a type of molecule found in natural substances such as chlorophyll. By contrast, Nantero, of Woburn, Mass., will store charges in arrays of carbon nanotubes.

Those are two of some 1,200 nanotech startups in the U.S., according to the latest figures from the NanoBusiness Alliance. They range from Nano-Tex, which makes the stain-resistant fibers for pants, and Zyvex, a Texas startup that supplies tools, structures, and materials to nanotech companies, to nano-pharma companies like NanoCure, which is researching nano-scale cancer treatments, and bio-nanotechs like Quantum Dot (see preceding Cool Companies article), whose products have wide applications in medical diagnostics and drug discovery.

The dazzle of the startups makes it easy to overlook the role being played by venerable giants. For example, nanotech has taken root at Cabot Corp., a $1.8-billion-a-year chemical outfit that is more than a century old. The Boston company makes carbon black--an ultrapure form of soot used as a pigment or reinforcing agent--and other substances. Over the decades Cabot has controlled the size of the particles that make up its chemicals with ever more precision, says Steven Reznek, vice president of R&D. "And in the past ten years," he says, "we've made nanoparticles." They are used, for example, to provide high-gloss coating on certain types of paper.

Such grown-up companies bristle at the notion that creativity belongs to the young. Says Williams at Hewlett-Packard: "I go to conferences and hear, 'Of course, small companies are the ones that have the most innovation.' And I'm thinking, 'Oh, please.' "

Strolling through HP's sprawling basement lab, beneath the preserved offices of the late founders William Hewlett and David Packard in the company's former headquarters in Palo Alto, the ponytailed Williams reveals what looks like a semiconductor clean room crossed with the lair of a mad scientist. Gleaming Rube Goldberg contraptions--which turn out to be things like a $500,000 scanning tunneling microscope that the HP scientists have customized themselves--are jammed everywhere. In the ceiling are rows of narrow pipes, built with bends to survive the motion of an earthquake, carrying an array of gases and chemicals. In the cubicles outside, some 30 chemists, physicists, and engineers trade ideas.

Williams' team has made breakthroughs in both nanotechnology and computer design. They devised a system that uses infinitesimal wires in a "cross-bar" pattern in such a way that molecules trapped between the wires become the equivalent of a bit of memory and/or a switch in a logic circuit. Their computing design eschews some of the highly expensive precision required in traditional chipmaking by creating an architecture that can tolerate a relatively high percentage of defects and still function smoothly. It's all, in Williams's view, the result of fundamental research that happens mainly when scientists aren't constrained by the fickleness of venture funding or the need to come up with a product quickly--HP has been bankrolling this group for nine years. "I look around, and the vast majority of the small companies are just hyping themselves like crazy--which they've got to do to make it. But in point of fact," Williams continues, "most, if not all, of the real advances have come out of large corporate laboratories." The giants, Williams argues, have the resources, the patience, and the experience to turn those advances into products. He cites IBM's GMR nanotechnology as an example. Williams goes even further: "The best way to kill a technology is to put it in a small company, because most small companies wind up going bankrupt."

General Motors is Williams's favorite example of how in innovation, size matters. GM's nanocomposites embody a unique combination of strength and flexibility. The company achieves that by altering the molecules of a clay so that it clings to an oil, which it wouldn't otherwise do. And though adopting nanocomposites for the trim on the Impala (and, as of May, on some Hummers) may not sound like much, GM plans to use 660,000 pounds of the stuff this year. According to Will Rodgers, a scientist at the company's R&D arm, GM will expand its use little by little in the next few years. "Right now we're dealing with trim. We'll be looking to move into the interior of the vehicle because there are some weight savings and some mass-saving opportunities. Eventually we hope to have exterior panels made from these materials."

Another old-economy stalwart that has joined the nano vanguard is General Electric. In recent speeches CEO Jeff Immelt has cited nanotech, along with diagnostic medicine and renewable energy (both of which are likely to be heavily shaped by nanotechnology), as "three big trends we're investing in for the future." The company has built an R&D staff of nearly 50 scientists and engineers working full-time on nanotech. "When we started the group three years ago," says Margaret Blohm, an organic chemist who heads the team, "we were kind of concerned, between you and me, 'Is it really real?' " Now, she says, "nano looks like one that's here to stay." In the end of 2003, GE began selling plastic, used in automobiles, with nanofillers that will allow paint to bind more readily to it. Blohm's group is studying myriad other technologies with potential applications throughout the company.

Nanotechnology is simply too important to be left to startups, says AtomWorks' Murdock. "This is going to be an innovation imperative for the FORTUNE 500 companies," he says. "Folks are going to have to know how to play in a world in which nanotechnology can have a very real impact on their competitiveness." And yet nanotech is often viewed through a much different prism. Complains Mark Modzelewski, who heads the NanoBusiness Alliance: "The nanotech race is getting caught up in the typical high-tech horse race aspects: judging it by venture-capital investments and the number of startups going public."

What we have here, clearly, is a technology that has gotten way ahead of itself. That's what explains the startup craze, and it's more obvious still in nanotech's sudden star power. Pop culture--which after all is the way most people learn about new technology--loves nanotech, especially its scary side. Michael Crichton's bestseller Prey, which is being made into a movie, features self-replicating flying military nanobots run amok. The preternatural powers of The Hulk's title character come from a nanotech experiment gone haywire.

Thrilling as the movies may be, they raise warning flags for business. With nanomachines still decades away, the specter of world domination by sci-fi-style nanorobots remains just that: sci-fi. But sensationalizing nanotech makes it harder to assess and manage whatever real risks the new technology may bring.

Take carbon nanotubes. These versatile cigar-shaped molecular structures are used increasingly in semiconductors, structural composites, and many other applications. To the naked eye nanotubes look like ordinary soot--you can order the stuff on the Internet for $50 a gram. But NASA-sponsored scientists have found evidence that nanotubes can cause lung abnormalities in mice. Researchers at Southern Methodist University and Rice University recently linked another type of carbon nanostructure known as fullerenes to brain damage in fish. Last summer, the Environmental Protection Agency declared ominously, "There is a serious lack of information about the human health and environmental implications of manufactured nanomaterials."

Will the growing use of nanotubes and fullerenes threaten workers or consumers? No one knows. But the uncertainties have already led some small environmental groups to oppose research on nanotechnology and to stage demonstrations outside labs. And if lab reports like those described above are treated as the dangers of nanotech rather than the dangers of a particular nanosubstance, nanotech companies, which so far have enjoyed the frisson of buzz attached to the word, could come to rue that moniker.

That, of course, is the pessimistic view. There's abundant reason to believe, based on the unexpected ways it's already showing up in our lives, that nanotech will continue to progress. But it is unlikely to unfold the way we expect, because science never does. The real Eurekas, says IBM's Theis, tend not to be visible until long after they've occurred. "When the really big breakthrough comes," he says, "it will not be evident to more than a handful of people."