Flash Memory Passes A Crucial Test

Several months back, in my cover story 'Song Wars: Striking Back Against the iPod Empire', I postulated that 'when a 4-Gbyte flash-based player costs the same as its equivalent-capacity hard-disk-based alternative, why would customers choose the bigger, bulkier, less rugged, and more power-hungry option if that's all the capacity they require?' Yesterday, Apple put that premise into practice, with the announcement of the flash memory-based iPod nano, and the coincident obsolescence of the 1" HDD-derived iPod mini. Nit-pickers will say 'not exactly', and they're right….the 4 GByte and 6 GByte iPod minis cost $199 and $249, respectively, whereas their equivalent-priced iPod nanos offer 2 and 4 GBytes of capacity.

It's not clear, however, how much of this discrepancy is due to actual bill-of-materials cost differences between HDD- and flash-based players, versus Apple's desire to extract additional profit margin from customers courtesy of the iPod nano's incremental feature set. And what an increment it is. The iPod nano is pencil-thin (literally), half the thickness of, and 62% smaller than, the iPod mini. It delivers a claimed 14 hours of battery life, nearly twice that of the original iPod mini (which claimed 8 hours, whereas the 2^nd-generation iPod mini touted 18 hours via a conversion to a next-generation PortalPlayer CPU and more aggressive HDD power management). And, unlike the iPod mini, it pulls off this battery-sipping feat with a color display.

In his January Storage Visions conference presentation, Semico analyst Jim Handy pointed out that, as my June article states, 'a hard-disk drive has a capacity-independent fixed cost derived from platters, heads, motors, chassis, and the like, whereas semiconductor memory's cost is more linearly related to capacity.' That idea's not new; you can find it in Toshiba's original mid-80's NAND flash memory presentations, in my presentations from the Intel days in the early 90's, and even in my book. What changes over time is the density threshold at which, putting aside flash memory's other benefits over rotating media that might prompt a magnetic-to-semiconductor conversion, its aggregate cost dips below the fixed cost of a HDD.

The notable aspect of yesterday's announcement, to me, is that we've now reached the point where a few GBytes' worth of flash memory is cost-competitive with a small form factor HDD. That's a key milestone. Because a few GBytes worth of flash memory stores a lot of tunes, particularly if you're on a server-archived subscription service and don't therefore need to tote more than a thousand or so tracks with you at any particular point in time. A few GBytes' worth of flash memory also represents a whole lotta JPEG pictures. See where I'm headed with this train of thought? Naysayers will point out that for an additional $50 you can bump up your iPod's storage capacity from 4 GBytes to 20 GBytes. And they're right. But $249, like $199, is a critical price point that motivates consumers (who'd fence-sit at $299) to pull out their wallets. And Apple can't (or won't) get there with a HDD-based iPod.

The same thought process applies to advanced cell phones, PDAs, digital cameras and the like. Going forward, the 1.8" HDD suppliers can cultivate their fortunes by cannibalizing today's 2.5" drive business (which is predominantly in notebook PCs). But the 0.85" and 1" HDD folks? They'd better hope that portable video capture, storage and playback takes off, big-time. Otherwise they're going to see their brief moment in the sun slowly but surely fade away, paced by the rate at which flash memory suppliers can build additional manufacturing facilities and execute lithography shrinks. Mark my words, small form factor HDDs are in trouble. You heard it here first.