System and component prices started their free fall during the bubble, when insatiable demand cleared shelves as quickly as they were stocked.
Price became an important competitive tool. Sometimes it was the only differentiator between two brands of hard drives or desktop PCs. When the economy dipped, prices kept falling as vendors tried to clear backed-up inventory and to prod customers into spending again.
Technology buyers are now spending again, but we can't call our present condition growth. Considering how far vendors' profits have slipped, no one's throwing parties over the sparse beginnings of the recovery. No thrills can currently be found in core IT markets such as desktops, notebooks, and PDAs. Basic rack-mount servers are a tough way to get by. Now, as before, price is the differentiator.
But there are a few exceptions to the not-for-profit conditions most system vendors face. If your IT organisation puts out a bid for a high-performance computing purchase, don't be surprised if vendors' CEOs line up to pitch their products. Individual rack servers don't make the registers ring; those boxes are now in the retail channel. But if you stack servers seven feet high and single-source their fibre switches, networked storage, and whatnot, suppliers will send up fireworks and construct buildings in your name.
Hard drive vendors were initially reluctant to pump scarce R&D dollars into SATA (Serial ATA) technology. Now SATA is quickly ascending to its intended slot: between ordinary ATA (standard desktop and notebook drives) and SCSI. SATA's on-drive circuitry is cheaper to build than SCSI's. Integration in mass-produced systems is easier, with no terminators, no jumpers, and no device numbers to sort out.
What may have surprised IT is the rapidity of SATA's uptake in networked storage. Midscale arrays, such as Apple's 2.5TB Xserve RAID, derive substantial price and margin advantages in the switch from SCSI to SATA. Customers don't care — it's impossible to tell SCSI from SATA when it's buried inside a well-built chassis. The lower end of the storage array market is still a competitive area. But vendors' finally have a way to reduce array-production costs enough to compete on selling price and still make a little money.
For me, the biggest surprise of the budding recovery is the location of system vendors' pot of gold. After years of being a tolerated (but sidelined) class of buyers, vendors are giving so-called "enthusiasts" the red carpet treatment. Why? Because these buyers — it's tempting to call them kids but they cross all age ranges — are great for business. They insist on the fastest processors, complaining loudly if Intel or AMD misses a planned ship date or has insufficient stock. When they buy complete systems, those systems are fully loaded: Maxed-out RAM, disk and 3-D graphics, all of which are highly profitable add-ons for system makers. When enthusiasts buy components to build or upgrade their machines, they're less likely to purchase in the buyer's "sweet spot" just below the state-of-the-art, where the bulk of consumers spend. An enthusiast may live in a shack, but he or she will make sure the hovel is equipped with the fastest PC that exists.
The group that benefits the most from the enthusiast market is chipmakers. Through competition, CPU makers Intel and AMD have forced each other to practically give away their processors. Previously, there weren't many takers for their newest, most expensive CPUs. Now the enthusiasts are buying so very enthusiastically, vendors are restructuring their single- and dual-processor technology to embrace these buyers. It's remarkable that a class of buyer virtually invisible to IT is spurring so many changes.
Typically, IT doesn't care how vendors make money, or how much they make. It isn't buyers' responsibility to maintain the profitability of manufacturers. But it is profits, not raw income, which will spur a meaningful tech recovery. For the economy's sake, I hope vendors keep looking for new ways to climb out of the low-margin pit they're in.