Posted by: Peter Burrows on November 1, 2005
Soon after I got back to the office yesterday after interviewing executives at Sun Microsystems, Dell Inc. pre-announced bad news for the second quarter in a row. That got me thinking: Could it be that poor old Sun, everyone’s favorite former thought leader, was right after all? For years, Dell has gobbled market share by applying its efficient ways to all corners of the PC business, while Sun has struggled not only for profits but for credibility in a market that didn’t seem to value its high-end innovations. But Sun chief technology officer Greg Papadopoulos made a compelling case that the pendulum is beginning to swing back towards Sun’s view of the world.
Essentially, he argues that computing is going through a bifurcation in systems design that will throw a kink into Dell’s low R&D approach to product development. Until now, he notes, all manner of PC-related products, from home PCs to high-end servers, were developed around the same set of Wintel industry standards. It was a “happy coincidence” that advances in desktop PC designs tended to increase the efficiency of Wintel servers, while technical advances in those servers tended to make their way into garden variety PCs over time, as well. “Everything that Intel did, just helped build a bigger and bigger business,” he says.
Now, that's starting to change, he says. The reason--and here's where the Sun "the-network-is-the-computer" schtick comes in--is that most of us are doing more and more of our computing out on the Internet rather than inside our own computers. Sure, we still write Word documents, use spreadsheets and create PowerPoint presentations. But more and more of our digital needs are met from some server farm out on the Net--when we request a search, download a song, send an e-mail via a hosted service such as Google's G-Mail, or use some Netified business application such as Salesforce.com. Indeed, it seems even Microsoft sees this trend.
What's this got to do with bifucation in system design? Because having such a smart network means the client devices don't have to be nearly as intelligent. That means that more of us will be using highly commoditized devices, designed to do a few particular things well while hitting low, low prices. That requires more inventive product planning, and the winners will be those that conceive of products with the right combination of size, features, and style--not those that wait for industry standards to emerge and do the best job of implementing them.
And by Papadopoulos' way of thinking, Dell also may face increasing issue at the other end of the market--for the Internet infrastructure gear that resides in those server farms. True, Dell has made a killing in recent years as many Net companies chose to string together gobs of low-end one and two processor Dell servers, rather than buy much pricier UNIX servers or mainframes. But as Google, Yahoo and their ilk dream up more and more services to push over the Net, they're going to need new kinds of gear. Otherwise, they'll need to build data centers the size of Rhode Island, cooled by acres of fans. Indeed, Sun execs note with glee that Google's single largest operating expense is its electricity bill, in part to air condition its watt-chomping server farms.
Now, I realize this is all classic Sun-speak--a self-serving vision that centers on the need for companies to buy its high-end gear. And of course, the market for basic Wintel and Lintel servers will continue to be a vast, growing one.
But truth is, the fastest growing part of the server market has been for blades--a technology that has not yet fully standardized and still requires plenty of proprietary systems management software and other types of R&D. And guess what? Dell has been largely a no-show. And there are plenty of signs of renewed innovation in the server space. Start-ups such as Azul Networks and Netezza are selling entirely new types of systems. And one of these days Sun will announce servers designed around its long-awaited Niagra chips. These "throughput computers" are designed from the ground up to help big Net companies and others handle gazillion small transactions at once--all those search requests, eBay transactions, and song downloads.
Of course, Michael Dell has figured his way past tough spots before, and Sun is still slogging its way through a gigantic one. But Sun is a company that's been counted out many times before, only to emerge stronger than before thanks to an idea that was ahead of its time. If Papadopolous is right, that time could be approaching again. What do you think?