Every so often some industry pundit comes along and announces that the PC is dead — soon to be replaced by Web tablets, terminals, PDAs or any of a host of other lightweight consumer devices. Over the years, companies like Gateway, Dell and Compaq have each announced the impending extinction of the personal computer as we know it. And every year, manufacturers roll out another new generation of PCs, each one more powerful than the last.
Putting it mildly, the rumors of the personal computer's demise seem somewhat exaggerated. The PC isn't dead. The major systems manufacturers just wish it were.
The problem is the ever-shrinking margins of the PC hardware business. Just because desktop processors get consistently faster, that doesn't mean each new engineering hurdle gets any easier. In fact, producing these continual improvements is extremely difficult, not to mention costly. Meanwhile, as the economy has declined, sales of expensive new PCs are down across the board.
Intel is already struggling to interest customers in its latest generation of chips. It's been slashing volume prices across its entire line, particularly with its most recent chip, the Pentium 4. But given its high cost and less-than-stellar performance, many PC manufacturers have opted to continue with the older Pentium III, rather than build systems around the new processor.
This reluctance to adopt its high-end consumer chips doesn't bode well for Intel's future enterprise direction, the superpowered Itanium architecture. Also known as IA-64, Itanium is an advanced processor design intended to compete with the likes of Sun's UltraSPARC, SGI's MIPS, Compaq's Alpha and other high-end chips.
But with this level of performance comes a trade-off: Compared even to the Pentium 4, Itanium is chips are phenomenally expensive. Savvy corporate IT managers see little reason to switch to IA-64 in favor of existing consumer chip designs — especially considering that very little software exists that takes advantage of Itanium's 64-bit architecture.
Instead, cost-conscious customers have preferred to build server systems using multiple consumer-grade CPUs. A popular choice for this practice has been AMD's Athlon series of Pentium clone chips. Athlons can run the same software and operating systems as Intel chips, with one major advantage: they're multiprocessor capable. Having two Athlons work together in a single box can produce overall performance comparable to a single 64-bit processor, for considerably lower cost.
The Pentium 4, on the other hand, can't be installed in a dual-processor system. It comes one per customer. And it'll be a while yet before Intel releases a version that's multiprocessor capable — if it ever does. It may prefer to concentrate its engineering efforts on the Itanium line.
If you think AMD's multiprocessing advantage and low prices will let it steal the market-leadership position away from Intel, however, think again. AMD and Intel have been engaged in a bitter price war over their desktop chips, one few analysts believe AMD will be able to sustain for much longer.
Past attempts to compete with Intel's deep pockets have failed, as well. Take the PowerPC processor, for example. The product of an alliance between Apple, IBM and Motorola, the PowerPC made a strong showing in the early 1990s with a series of fast chips, and for a brief period actually managed to outpace Intel's Pentium line.
But with Apple as the only volume purchaser of PowerPC chips for desktop machines, the financial support just wasn't there to compete with Intel and its incredible rate of processor speed increases. Eventually the Pentium pulled ahead, and today the fastest PowerPC chips run at only slightly more than half the speed of Intel's top-of-the-line consumer CPUs. While Apple argues that its G4 PowerPC offers "supercomputer performance" to rival any chip on the market — even at the lower clock speed — that argument can be a hard sell with skeptical consumers.
But continued market dominance can be seen as a Pyrrhic victory for Intel. The end result of years of playing hardball with processor pricing is what we have now: a desktop PC market where the bottom has all but dropped out. Consumers have come to expect ever-increasing CPU performance, while at the same time demanding the lowest possible prices for their PCs.
Ultimately, building personal computers is a losing proposition. With margins as low as they are today, hardware manufacturers have little choice but to pray for a shift in the computing market — one that will get the PC out of their hair once and for all.
What direction might that shift take? The case of Transmeta may be a telling one. Some analysts hoped that the launch of Transmeta's much-hyped Crusoe chip in January 2000 would herald a new era for PC processors; unfortunately, so far the reality hasn't lived up to that hype.
The high hopes for Crusoe stemmed from its innovative architecture. Unlike AMD's Athlons, Crusoe isn't exactly an Intel clone chip. Instead, Crusoe uses an ingenious translation process that dynamically rewrites code intended for Intel processors. The translated code will then run efficiently on Crusoe's own architecture while drawing much less power — or so Transmeta has claimed.
Crusoe's real-world performance has been disappointing, however. So much so, in fact, that some Transmeta shareholders have launched a class-action lawsuit, claiming that the company's executives knew its chips wouldn't meet initial performance and power-consumption claims.
But the Crusoe chip was never really intended to compete with Intel's high-powered desktop CPUs. Instead, Transmeta was hoping to sell its processors to manufacturers of low-power connected devices, like notebooks and Web-enabled tablets — devices that would, it was foretold, replace the PC.
In part, this vision can be traced to the software market. In the new model for the software business, which has been embraced by Microsoft and Sun, among others, the processing power for a new generation of applications would reside chiefly on servers hosted somewhere on the Internet. Rather than owning a copy of a software application, customers would "rent" it, downloading the latest interface code to some kind of lightweight client, like a Web browser.
User data would similarly be stored on networked servers — no hard drives necessary. Users could log in from any networked location and still have access to their documents, contacts and preferences.
The idea is that the future of the software business lies in offering software as a service, rather than as a packaged product. Companies like Microsoft don't want to sell software to their customers just once; they want those customers to subscribe and thus keep paying every month, whether they want new features or not.
It's a questionable idea for consumers, many of whom have concerns about not just cost but the privacy implications as well. It's hard to deny that such a model would be great for software companies, though — and it would be great for hardware manufacturers too. Just think: In a world of low-power, handheld, "dumb terminal"-type network devices, no longer would chipmakers be forced to concentrate on the eternal march toward increasingly higher-powered processors for the consumer market.
Instead, they could concentrate those efforts on the server market, where well-funded corporations would be willing to buy in to the big tickets of products like Itanium. On the consumer side, they could then channel their efforts into streamlining the manufacturing process and producing smaller, less expensive chips.
That's the world Transmeta was hoping to get a running start into. Let's just say, however, that the market for connected tablet-type devices didn't mature quite as soon as predicted.
In fact, who's to say it ever will? So long as low-cost commodity PC hardware can do everything a connected tablet can do — and more — consumers have little incentive to switch to a method of computing that's radically different from what they're used to.
The ultimate question is, Just because companies like Microsoft and Intel would like to see a change in the marketplace, does that mean it has to happen? Is it possible that the personal-computing industry has made its bed and now must lie in it? Or do we, the consumers, really have no choice in the matter?
My advice is to keep an eye on your PC these days. Some companies out there would like to see it dead — and they're used to getting what they want.
