(FORTUNE Magazine) – For the past decade, the chip business has been relatively easy to understand: There was Intel, and then there was everybody else. But a couple of recent developments are changing that hierarchy. First, for the past three years the growth of chips devoted to handling communications--Internet access, data transmissions, wireless and land-based phone calls--has exceeded the growth of chips designed for PCs. Cell phones, routers, set-top boxes, and modems are swallowing up more and more silicon.

The second development is Furby. Yes, Furby.

The little gremlinesque toy and its relatives, like Rugrat Baby Dil Pickles, are powered by chips that fit more and more features onto a single sliver of silicon. As the inner workings of communications and consumer devices begin to fit on a single chip, the $120 billion industry will take on a different look. Intel, which sold $26 billion of chips last year, no longer has a stranglehold on the future. Second-tier giants such as Texas Instruments, Motorola, and Lucent Technologies are battling to rule new and potentially lucrative territories, and newcomers such as ARM Holdings, Broadcom, and PMC-Sierra are carving out their own niches by doing business in a brand-new way. Bob Bailey, CEO of PMC-Sierra, which makes chips for high-end telecom and data switches, explains, "What you are seeing is a dramatic restructuring of the semiconductor industry."

If you examine these trends closely, it's easy to understand why. Before the Internet, PC processing power couldn't keep up with the appetite of bulky software programs. Now it can, and instead chips that deliver more bandwidth are in demand. Integrated Circuit Engineering, a market research firm in Scottsdale, Ariz., forecasts that revenues for communications chips will grow 15% annually over the next five years, vs. 10% annual growth for computer chips. Chips going into computers still account for roughly half of total industry revenues, compared with about a fifth for communications chips. But as Merrill Lynch analyst Joseph Osha suggests, "The growth rate in PC chip consumption may have topped out."

As more and more services reside on the Net, non-PC devices built to handle communications--such as cell phones, set-top boxes, and Palm-type gizmos--will become virtually ubiquitous. The International Data Corp. research firm thinks devices like the Palm will become a primary way people access the Net. IDC researchers believe that the number of such devices will rise at a 76% annual rate, from a scant six million last year to 56 million in four years. Meanwhile, the number of PCs is expected to grow at only a 13% annual rate, from 89 million to 146 million worldwide.

So how does Furby fit into all this? The little $30 doll is the latest in a line of products, like videogames and digital home- entertainment systems, that herald the merger of computing and consumer electronics. It shows that pulling the functions of multiple chips onto a single one (called "system-on-a-chip" in industry parlance) makes possible a whole range of new digital devices. A Furby is controlled by two chips. Two years ago, the plush monster would have required at least four--and have cost twice as much. Furby's chips make it giggle, snore, learn words, and react to outside stimuli. If your child does not pretend to feed her Furby, for instance, the doll will catch a cold, and soon infect other Furbys within receiving range.

Furby and its ilk are just the start. Companies are making systems-on-a-chip for cell phones, set-top boxes, and DVD players. The ability to do so is a direct result of Moore's law, which predicts that the number of transistors that can fit on a chip doubles every 18 months or so.

The resulting superabundance of transistors has always allowed companies to integrate multiple functions on a chip. But now everyone is trying to master the art of mixing and reusing different blocks of circuitry to create amazing new products. Lucent, for instance, will be able to make a cell-phone-on-a-chip next year that combines 11 chips into one. The $30-billion-a-year company sold $3 billion of communications semiconductors last year; more than half that revenue was from systems-on-a-chip.

Lucent faces tough competition in communications semiconductors. Over the past two years Texas Instruments has phased out much of its electronics business, and it is now focused mainly on manufacturing special digital-signal-processor (DSP) and analog chips, key ingredients of the communications revolution. Intel just formed a join venture with Analog Devices to make DSP circuitry blocks. And communications juggernaut Motorola is building standardized circuitry blocks that can be shared among its many divisions. IBM is following a similar strategy, offering its customers a library of common reusable circuitry.

Younger companies are also making a splash. For instance, Broadcom in Irvine, Calif., holds a Microsoft-like market share in chips for cable modems. With a lock on part of the telecom switch market, PMC-Sierra has higher gross margins than Intel. And there are even "chipless" chip companies that do nothing more than sell their ideas. ARM in Cambridge, England, designs small microprocessors around which other companies can build larger chips or, say, an entire cell phone. Every chip company would like to be the Intel of the communications era. But that's highly unlikely. One reason is that these chips will be used across disparate markets. Jack Kilby, co-inventor of the integrated circuit, points out another: "There will be no company that has all the blocks [of circuitry] someone might want." Not even Intel itself. Not surprisingly, some of the folks at Intel have a hard time envisioning a world in which companies swap intellectual property as they assemble new kinds of chips. Albert Yu, who oversees Intel's design efforts, says, "The utopia is that everyone can pull from intellectual property to which everyone else has access. But where is the value in that? Who is in control?"

Losing control is perhaps Intel's worst nightmare. But it's already shown that it can react quickly to threats to its hegemony. As it seeks a place in the emerging information-appliance market, Intel seems to be expanding beyond its Pentium-only policy. It sells a low-end chip originally based on an ARM design (dubbed StrongARM). And it is working with Mattel to make toys that are an extension of the PC. What's more, the growth of communications benefits Intel, since its chips power the PCs and servers that manage Internet content. So there are good reasons why Intel is still mostly obsessed with continuing to dominate the PC market, where semiconductors typically cost $200, vs. less than $40 for a chip that goes into a communications device. Your kid's Furby may giggle and burp, but it isn't yet singing a dirge for Intel.

INSIDE: Look at the stuff $6 billion can buy... Why Y1999 will be a nuisance... Shaffer on Web robots... Tonight on ET: A telecom exec goes Hollywood... The Dreyfuss Report on cleaning up your hard drive... Alsop on absurd Net stock valuations