Ten bets for HDD to SSD transition

Ramdas S

I believe Western Digital acquisition of SanDisk announced last week is the first real signal that SSDs are going to replace magnetic spinning drives very soon.

Western Digital which has revenues of about $14.7 billion is one of the three remaining manufacturers of spinning magnetic drives commonly called hard disk drives. Its biggest competitor is Seagate, with Toshiba being the third independent brand in HDD manufacturing globally. Western Digital which had acquired Hitachi’s hard drive business that runs that business as an independent subsidiary called HGST.

Over the past three years, the magnetic drive market has slipped primarily because of a drop in the PC business. At the same time faster SSD drives have seen steady growth rates in volumes.

However SSDs are still expensive. A 1 TB Seagate internal HDD is retailed on Amazon India for Rs 3,770 while a similar capacity Samsung 850 EVO 1 TB 2.5-inch SATA III internal SSD is priced at Rs 29,573. This represents almost 7.8 times price difference. The numbers are little better when you compare a capacity such as 250 GB where the difference is at best about 3 times.

Over the past two decades data boom helped disk drive industry grow and magnetic disk drive industry worked on their own Moore’s law by doubling capacities at sweet spots of $55 or about every 2 to 3 years. With software vendors shipping software whose footprint grew exponentially, there was always an urge among consumers to keep buying drives with larger capacities. Also consumers have been storing high resolution images, videos and music and have been filling up disk drive capacities.

However like what happened with Moore’s law for microprocessors, magnetic disk drive markets have hit a wall of sorts. Most OEM vendors have so far been unwilling to ship mainstream PC models with 2 TB hard drives despite the price difference between 1 and 2 TB being just Rs 2,000 or about $30. In fact there is not a single notebook which is shipping less than Rs 60,000 featuring a 2 TB drive.

There are fundamentally three issues behind OEMs being reluctant in shipping larger capacity drives.

1. Most PC users do not want to store high amounts of data in their primary drive. Most of them prefer using an external drive, cloud storage or NAS device for storing larger capacities.
2. As capacities grow, magnetic drives become slower as it requires more time for spindle to spin and hence data access times are much lower. Hence few users really prefer to run a magnetic drive more than 1 TB as primary drives.
3. The disk drive industry’s experiment to ship hybrid drives which are essentially drives with a small portion of the capacity served with flash while rest of the capacity served through a magnetic drive has met with lukewarm response so far. Some of the models were known to have high failure rates, have also not inspired the buyers.

SSD drives with no moving parts offer read and write speeds excess of 500 Mbps and write-reads between 25,000 and 100,000 IOPs. Unlike HDD drives larger capacities don’t slow down SSD drives. While reliability issues plagued SSD drives, leading manufacturer OCZ, a Toshiba company has published a report that claims reliability has improved to the magnitude of 40 over past 2 years.

Will HDDs die?
Eventually it will, but not for the next 5 to 7 years. HDD manufacturing despite heavy automation is a highly labor oriented process. With several moving parts there are over hundreds of suppliers across the Far East, especially in Thailand which drives the economics of this industry.

An immediate fallout of potential drop in volumes could upset economies of scale, further squeezing potential margins enjoyed by the industry, which is already single digits for capacities less than 2 TB. With current costs no OEMs will be able to manufacture hard drives at less than $ 40 and still continue to sustain the business.

SSD drive manufacturing on the other hand is very similar to memory module manufacturing with majority costs being that of the silicon and the drive controller. With no moving parts the industry is dependent on the large semiconductor manufacturers and designers such as Samsung, Intel, Micron, SanDisk, Kaminario, Hynix, Extrememory and Toshiba who makes solid state memory. There are also about a dozen vendors who make the controllers including names such as Toshiba, Intel, Samsung and SandForce. At this point of time there are over 40 recognized SSD drive manufacturers globally.

Key players such as Intel, Samsung and Hynix have announced billions of dollars of investment in new manufacturing fabs that can create SSD drives with capacities in excess of 10 TB. Samsung has already announced a 16 TB drive priced at $8000. It has a 2 TB enterprise class drive retailing less than $ 1000 with a 10-year warranty already shipping.

Based on available data, I am willing to make some bets on where the SSD and HDD industries are headed, along with a timeline when certain breakthroughs can be expected.

1. I am expecting entry level consumer 240/250/256 GB SSD drives to retail at $65-70 this Christmas buying season. Currently prices are averaging less than $80. This price point is significant, as a 1 TB 7200 rpm HDD will be priced at about $ 55 around the same time. Performance hungry consumers would consider picking up these drives for upgrades despite 4 times bigger capacities being available on spinning drives.
2. I am also expecting the first models of notebooks with 240 GB SSD priced for the sweet spot markets to emerge in another 3 months which will be priced at $599 or less or about Rs 44,000 in India. Currently the lowest priced model which ships with an SSD for Wintel market is an Asus Zenbook at Rs 53,000 on the street. Asus Zenbook is based on the Intel ultra book specification.
3. In first half of 2016, I am expecting non ultra book models or vanilla notebooks shipping with 250 GB and 500 GB SSDs which would be priced at sub Rs 40,000. The entry level 500 GB consumer SSD prices should fall to $110 or about from the current $150. This would trigger launch of a number of sleek notebooks which may not conform to Ultrabook standards but would still offer a great alternative to a value conscious power user. The price points are possible if the Dollar-Rupee maintains a stable tango could be as low as Rs 30,000 by end of 2016. Yes, expect an advertisement for a Core i3 notebook retailing at Rs 29,995 with an SSD drive of around 500 GB capacities by December 2016.
4. With larger OEM interest in H2 of 2016 we should witness further drop in SSD prices. The 240 GB SSD would retail at $ 40 or about, while the 500 GB SSD will be available at less than $ 70 by Christmas 2016. This would increase SSD shipments to a little over 40 percent of all notebook shipments, and accelerating the transition.
5. I am expecting more SSD drives shipping than hard disk drives in 2017. Digitimes Research agrees with me on this count and the prediction is 54 percent of drives will be SSD in 2017.
6. Hard drive makers will focus on larger capacities, so expect the per TB price on HDDs to drop to as low as $ 15 per TB in 2016 and $ 7 per TB in 2018.  I won’t be surprised if a 5 TB HDD is shipping as low as $ 75 in 2016. However magnetic drives would largely be seen in external drives and SME class NAS devices than in notebooks or desktops.
7. An immediate fallout could be further consolidation, and Toshiba could be a target for acquisition likely by Seagate. Seagate could also buy one of the Flash memory manufacturers to protect its industry leadership in future too.
8. Magnetic drive industry will start focusing on building larger capacity drives targeting external drives and for secondary storage, shifting away from primary storage. Cloud and DC backup vendors have shown a lot of interest in larger capacity to create archival services. The secret sauce for survival for traditional hard drives could be a form laser technology, known as heat-assisted magnetic recording (HAMR), which is able to write smaller, more-stable bits onto the magnetic surface of a spinning disk. Today, Seagate's largest capacity drive using conventional technology is 10TB. Using HAMR, that capacity could theoretically increase to 30TB.
9. Once 3D NAND technology which Samsung is championing takes mainstream, magnetic disk drive makers will be under even more pressure. By 2020 a NAND Flash drive of 30 TB capacity would be cheaper than a 30 TB spinning drive, which could very well kill the magnetic disk drive market. By 2017, you will see first of the external disk drives sporting an SSD drive inside.
10. Enterprise disk drive market will migrate completely to Flash by 2017, except for big data storage and some file based storage workloads. Database workloads would move on to Flash completely. Presently the price equations are such that SSDs are about 4 times more costlier compared to 15,000 RPM SAS drives and over 10 times over enterprise SATA. But even a price drop by 50 percent over next 12 months would see a number of customers dropping traditional SAS and SATA based HDD drives in favor of SSD. With public cloud vendors such as Amazon, Google and Microsoft offering SSD even for entry level cloud instances, the server vendors would start shipping SSDs as standard in 2016 even for entry level servers.

Caveat: These predictions do not take into consideration potential acts of God and natural or man-made calamities.

The article is authored by Ramdas S, CEO, Netzary