'

SSD prices: how low will they go?

If you've been holding off on buying a new SSD or high-capacity thumb drive due to price, there's good news: prices are dropping fast. The capacity squeeze that sent NAND flash and DRAM prices soaring last year is ending -- but a new bottleneck is looming.

Samsung Electronic's new SSD: 33.2 TB of memory for enterprises and datacenters

I'm a big fan of SSDs in notebooks. If you have an older laptop -- 2012 or earlier MacBook for instance -- replacing its hard drive with an SSD is a no-brainer upgrade that will add years to its life. Same with Wintel notebooks, of course, but usually you'll be better off buying a new model and getting the latest bells and whistles.

special feature

The Evolution of Enterprise Storage

How to plan, manage, and optimize enterprise storage to keep up with the data deluge.

Read More

But starting in late 2016, a weird thing happened: after declining for years, prices reversed and started climbing. This was partly due to climbing smartphone capacities and sales, and partly due to a pause in bringing new capacity online.

So prices spiked and I, for one, stopped buying SSDs. Of course, Apple and Samsung, the two largest consumers of NAND flash, didn't. They weren't paying retail either.

But in late 2017 new capacity started coming on line. And the migration to three-level (TLC) and quad-level (QLC) flash gained traction, increasing effective capacity.

So supply and demand started to equalize. Now flash prices are expected to decline 15 percent this year on a 45 percent increase in supply.

DRAM is a different story, with prices expected to climb again this year by another 23 percent. Mobile is the driver for DRAM prices, as it is for flash.

The new bottleneck

The threat now is not fab capacity. It's wafer capacity.

Chips are built on wafers. Different kinds of wafers are optimized for the production of various semiconductors, such as discrete components, or, at the other end, microprocessors.

Fabs buy wafers on the open market. And today, the companies that produce wafers are supply constrained.

As long as fabs are able to pack more capacity on a wafer, the market balance will remain. But if demand pushes production beyond what the wafer vendors can provide, we're looking at higher prices.

The Storage Bits take

Given all the heavy breathing around "trade wars" and tariffs, one would think that our supply chains look like board games. Move the pieces around, win -- or lose -- some territory, and move on.

But supply chains are a lot more interwoven than they were 20 years ago. Just-in-time means no-buffer-stocks. So when supplies tighten -- or are interrupted -- as they were with the Thai floods, the effects are near instantaneous.

The NAND flash and DRAM pricing issues of the last couple of years underscore how sensitive international supply chains are to disruption. Industry execs understand this. Unfortunately, not everyone in Washington does.

So buy those SSDs, thumb drives, and DRAM now. We could be in for a wild ride.

Courteous comments welcome, of course.

Previous and related

PGP encryption won't protect your data. But PURBs can

You may think that encrypting your sensitive files with, say, PGP may protect your data -- but you'd be wrong. Most encryption formats leak a lot of plaintext metadata, and that's a problem. Here's what you need to know.

Can this 100+ TB cloud SSD controller herald the all-flash data center?

A decade after flash burst on the enterprise scene, all-flash data centers remain the exception. But startup Burlywood plans to rearchitect flash SSDs so even hyperscale data centers can go all flash. Here's how.

Why SATA flash drives are being left in the dust

The SATA interface on SSDs was a marriage of convenience, not love. Flash SSDs are so fast that SATA simply can't keep up. But NVMe can, and that is changing how how systems are configured and infrastructures are designed.

Fail-slow at scale: When the cloud stops working

Computer systems fail. Most failures are well-behaved: the system stops working. But there are bad failures too, where the systems works, but really s-l-o-w-l-y. What components are most likely to fail-slow? The answers may surprise you.