Western Digital is focussing more on HAMR as its breakthrough technology to into the 30TB and beyond disk drive area, with no flash-disk cost crossover for at least ten years.
This became clear with a CEO Dave Goeckeler session at the virtual Wells Fargo TMT Summit 2021 that took place on November 30.
He said the HDD market looks to be in a period of sustained high demand. “Post-COVID … people [are] using the cloud more and more all the time … [and[ the cloud [is] powering ever more intelligent devices connected by high-speed networks. We play in two very, very important parts of that ecosystem and that platform — that is storage in the cloud and storage on the device.” This is “driving good demand across the endpoints and across the cloud.”
Goeckeler said that the incredible expansion in unstructured data storage by businesses and other organisations was a long-term tailwind for disk drive manufacturers. “We’re entering a stage where the cloud is just continuing to grow at a phenomenal pace. We’re all using it more and more.”
WD needs to keep innovating to lower the $/bit cost of its storage. Goeckeler prognosticated, “if we … continue to drive storage innovation and lower the cost of storage per bit, I think that we have a very, very strong demand driver in multiple dimensions; increased use, increased demand and also, there’s a big part of the iceberg below the water of data that’s not stored that could be stored with a different — if we can continue to drive the economics in the right direction.”
He said the bulk of the business was selling to enterprises, not clients, and that meant a change in internal investment emphasis to focus on “thinking about, well, how much do I need to invest so I can meet demand two or three years out?” So “we’re thinking about investment and being able to fuel that growth.“
The discussion took a look at Heat-Assisted Magnetic Recording (HAMR) technology which uses laser-produced heat to create smaller bits than conventional magnetic recording (PMR) that are stable at room temperature, unlike similar-sized PMR bits. WD has been emphasising the use on microwaves in its Microwave-Assisted Magnetic Recording (MAMR) as an intermediate step to HAMR and made an initial advance in that direction with its enhanced PMR (ePMR) now called energy-assisted magnetic recording (EAMR).
Goeckeler said: “HAMR is a very important technology. There’s no doubt about that. We’re heavily invested in HAMR. I think you know we have over 400 patents in HAMR. Any time you’re a supplier of hard drives in an industry this big, you’re going to be invested in a number of different technologies that you think is going to fuel your road map. So we’re a big believer in HAMR.
“I think the industry now is coming more around to the realisation that the HAMR is going to be real, it’s going to be in the future. It’s going to be very important. It’s going to extend the life of hard drives for a very, very long time
“We’re focused on HAMR, but we’re also focused on the steps to get there.”
The steps so far have been energy-assist and OptiNAND (flash-enhanced controller), about which he said it’s “very, very important technology, higher reliability, better aerial density. It allows us to … deliver several generations of technology. We’re able to deliver our 20-terabyte on nine platters, we can add the tenth, and we get another 2.2 terabytes of storage.”
With Optinand, “we really have that staircase to take you to 30 terabytes and then you get on the HAMR curve and you go for quite a bit longer. So I think it’s a really good story for — a really good road map for the hard drive industry.”
“Several generations” implies to us at least three generations of OptiNAND drives. If they each add 2TB then we are looking at a 20TB (nine-platter), 22TB (ten-platter) and then 24TB (10-platter), 26TB (10-platter) and 28TB (10-platter). Add a fourth OptiNAND generation and we are at Goeckeler’s 30TB doorway to HAMR.
We can derive a rough timescale from this, by assuming a 12-month cadence:
- 2022 — 24TB
- 2023 — 26TB
- 2024 — 28TB
- 2025 — 30TB
WD competitor Seagate is currently shipping first-generation 20TB HAMR drives and has a 30TB HAMR drive in development, with a cloud supplier market in prospect, not the traditional enterprise market. That will be served by conventional PMR (Perpendicular Magnetic Recording) drives.
Colin Presley, a Senior Director in Seagate’s CTO organisation, told us in September that HAMR “is really really hard technology” and “The industry across the board recognises HAMR is the road to high capacity.”
If Seagate produces a 30TB HAMR drive by, say, 2025 then that would be when WD is about to produce its first HAMR drive. By then Seagate will have shipped many millions of its HAMR drives and be vastly more experienced in manufacturing the drives, understanding their reliability, and creating controllers and firmware for using them.
We could see a situation emerging where WD is at a competitive disadvantage because it is years behind Seagate in HAMR manufacturing and controller technology.
Shingling and flash-disk crossover
Goeckeler said OptiNAND makes can be applied to Shingled Magnetic Recording (SMR) disks and they add additional capacity, say in the 15–20 per cent area. But “clearly, something like SMR requires changes on the client side. That’s not something anybody takes lightly.”
SSD and HDD $/TB prices are going down. Says Goeckeler: “They’re both declining,” and “We’re very comfortable with our [flash] road map and driving 15 per cent cost down there.”
He said: “But on the other side, you’ve got hard drives continuing to go deeper into the well of data that’s out there, and we see that — we see those costs continuing to go down. Like we said, we talked about a road map here that is many, many steps into the future with a major technology transition like HAMR in our future, it’s several years out, but it’s in our future. And so it provides a lot of runway into that drive value proposition.”
Ultimately he is not worried abut flash becoming cheaper than disk for the next decade. “Storage in the cloud is very important. The vast majority of that data is stored on hard drives and will be for a very, very, very long time. In fact, if you look at the economics, we don’t see crossover for beyond a decade, which is … beyond the planning horizon of any useful technology business.”