Could Western Digital Be Broken Up?
In late September WDC said that it had created two separate business units. One focused on HDDs and the other on flash memory. The company also announced that it hired a former Symantec and Cisco executive, Robert Soderbury as GM of the flash unit and that it was looking for someone to lead its HDD business. The company did not indicate that it planned to divest or spin off either of these businesses at that time, but there are analysts who think WDC’s CEO David Goecheler (CEO since March 2020) may be preparing for significant changes in the company.
Let’s take a look at the flash memory and hard disk drive (HDD) business at WDC and in the overall market and see if we can get some sense of what each of these businesses have been like over the last few years and putting long term trends into perspective, see if we can get some insights into what these two businesses would be like if they were independent.
Both flash memory and HDDs are digital storage devices. Flash memory is used in many consumer and industrial applications. The biggest consumer of flash memory is in personal mobile devices such as smart phones. As the price of flash memory has been declining, solid state drives (SSDs) made with flash memory have made serious inroads on what had been HDD applications, such as storage in personal computers (both desktop and notebook).
SSDs have higher higher read/write speeds than HDDs. They are also more rugged and shock resistant. HDDs are less expensive for a given storage capacity. Client HDD (such as in PCs) are around 5 times less expensive than SSDs for a given storage capacity. Very high capacity HDDs (often used in enterprise nearline storage in data centers) are about 10 times less expensive than SSDs for a given storage capacity.
The lower price difference for storage in client applications combined with slowing demand for additional capacity internal to these devices and the advantages of SSDs, have caused legacy client HDDs sales to fall, and SSD market share to increase. At the same time, the growing demand for data center storage, particularly in hyperscale data centers, has led to significant growth in high capacity nearline drives. As a consequence, although total HDD shipments at WDC have declined, the average sales price (ASPs) for HDDs has increased.
In fact, WDC HDD ASPs have increased by about 43% from C3Q 2017 through C2Q 2020, while WDC HDD shipments have declined by about 50% over the same period. As a consequence, WDC HDD revenue only declined about 21% over this period. This has resulted in fairly flat gross margins for HDDs at about 30% over this period. On the other hand, gross margins for WDC SSDs have changed significantly over the same period. How come?
Unlike in the HDD industry, which only has three competing companies (Seagate, WDC and Toshiba), the NAND flash industry has six major players (Samsung, Kioxia, WDC, Micron, Intel and SK hynix) and a few smaller manufacturers. In addition, the NAND flash fabs, particularly with the move to 3D NAND cost billions of dollars to build, compared to a few hundred million to build an HDD production plant.
In the NAND flash industry, when prices are high due to a shortage of production compared to demand, all the manufacturers build new production capacity, which generally takes 18 months to 2 years to come on line. Then all this new production capacity comes on-line at about the same time and the supply of NAND flash greatly exceeds demand and prices drop. With less revenue and profit, the NAND manufacturers don’t invest in new production facilities until demand increases and begins to exceed supply and prices and profits rise again.
With relatively steady increase in NAND flash demand and big swings in NAND production capacity we end up with a boom/bust cycle in the NAND flash industry. This has been going on in the industry for over two decades with the cycle time generally over a 4-5 year period. At present we are in an oversupply of NAND flash and prices are low, but possibly getting ready to recover. Note that the general cost of NAND flash per capacity declines over time and flash density increases, so the resulting swings in the price of NAND flash are swings about a declining trend line.
When there is an undersupply, like there was in 2017, NAND flash and SSD prices increase and gross margins and thus profitability, are high. A lot of 3D NAND capacity came on line in 2018 through 2019 and that drove NAND contract prices down from 2017. In general, undersupply in NAND drives higher prices for SSDs and thus better SSD gross margins.
There were signs that contract prices were increasing slightly in early 2020, but COVID has dampened demand and could extend the period of oversupply into 2021. If YMT from China begins to provide some meaningful production in 2021 that oversupply situation could be extended further.
So, what does this tell us about the viability of WDC’s flash memory and HDD
as separate businesses. Overall WDC has rebounded somewhat from a low point in 2019. During the latter part of 2019 HDD revenues were driving this rebound. In 2020 HDD revenues have decreased and flash revenues increased. The 2020 flash memory rebound is likely due to a combination of reduction of flash excess production capacity combined with some increase in demand.
In general, the WDC HDD and SSD business are pretty comparable in revenue (most times), so breaking up the company would create two smaller companies with relatively equivalent revenues (at least for a while). However, as more legacy HDD applications migrate to SSDs HDDs revenue will decline and SSD revenues will tend to increase.
At some point in the future, assuming that nearline HDDs continue to advance HDD capacities so they maintain about a 10X difference in price compared to enterprise SSDs, the majority of WDC’s HDD business will be in nearline (and other high capacity) HDDs. We project that this will occur between 2023-2024. At this point, the general annual decline in HDD shipping volumes will be much less because demand for higher priced high capacity HDDs will remain strong and revenues will generally follow enterprise and data center refresh cycles. HDDs will become mostly used in large storage installations and thus if WDC created a separate HDD company, its primary focus would increasingly be on these storage markets.
On the other hand, flash memory use in consumer devices will expand, creating greater demand. The need for high performance storage to support data centers, edge computing and enterprise applications will drive flash memory demand from that side as well. Thus, a broken out WDC flash memory business will experience continuing increase in demand from multiple user markets. On the other hand, if the number of flash memory competitors remains as it is today, the flash company will continue to experience the boom/bust cycle that has characterized the NAND flash industry, leading to large swings in profitability.
A flash memory focused company would do well look for opportunities to consolidate and increase its market share where that made financial sense. Fewer competitors would probably reduce the production oversupply during the bust periods and provide more revenue to help the company weather these periods better, while giving them funds to develop the next generations of products.
Breaking up WDC into a separate HDD and flash memory business could create two viable storage company, that over time will provide value to very different, but potentially symbiotic, uses. The HDD business will become enterprise and especially data center focused, while an NAND flash business will seek to gain market share as a way to moderate and better survive the NAND flash boom/bust cycles.