Why SSDs won't replace disks

It's simple: we don't produce enough flash to replace more than a small percentage of the capacity that disks provide today. But drive vendors would be foolish to take comfort in that fact. They are still in a world of hurt.

In a recent paper titled NAND: can meet the growing storage capacity demands of the laptop PC market?” (pdf) Seagate, soon to be the world's second-largest drive manufacturer, points out that the market for laptop PC disk drives worldwide is 69 exabytes (EB) and is forecast to grow to 95 EB in 2011.

But last year, the flash industry only produced 11 EB of capacity. And 90% of that went into consumer devices such as smart phones, ST cards and drives. It costs over $2 billion to build an exabyte of flash production capacity.

That means an investment of about $200 billion in megafabs to build enough flash to satisfy this year's market demand. Even if equipment vendors could build the fab lines, no one in today's world would finance it. It isn't going to happen.

So what will happen?

Two things:

• As the advantages of flash-based notebooks and tablets become more obvious to notebook consumers, people will start asking themselves how much storage they really need.
• SSD-based notebooks will take over the higher end of the market, where the margins are better and average sale prices higher.

This is the problem for disk drive vendors. Sure, people will keep buying low-end notebooks with disks. But the already brutal price competition will get worse and already thin margins will get thinner.

The Storage Bits take The main reason Seagate wrote the paper is to pimp their clever Momentus XT hybrid drives. Since we can't all have SSDs, hybrids may be the next best thing.

Seagate has a point: most of the data on consumer machines - most of the world's data on any machine - is rarely accessed. Why keep it on the most expensive storage if we don't have to?

But the larger point is worth repeating: flash SSDs won't replace disks because we can't build enough flash to do it - even if we were willing to pay the price. And the drive industry is far from done reducing storage costs, so they remain a fast moving target for every non-volatile memory technology.

Which is good news for the data hoarding masses.

Comments welcome, of course. Seagate sent me an XT to review and I found it a worthy effort for the reduced boot times alone. But I hope they are busy tuning their algorithms and doubling their flash capacity to improve performance over a broader range of tasks.