WD's new HDMR tech to enable record-breaking 100TB+ drives
Bit patterned media expected to arrive in the next decade.

Western Digital is set to introduce its first hard drives featuring heat-assisted magnetic recording (HAMR) technology, and around 2030, this technology will enable the company to build 80TB–100TB HDDs. At that point, FePt-based granular media will start reaching its limits in terms of areal density. But Western Digital is already working on the next step for its hard drives: heat dot magnetic recording (HDMR) technology.
The HDMR is a next-generation recording technology that involves heat-assisted writing and bit-patterned media (BPM) that promises to achieve areal densities of around 8 Tb/inch^2 and beyond (according to Seagate) and to enable 10-platter HDDs with capacities of 120TB or higher. This technology is projected to be quite expensive as bit-patterned disks must be physically patterned using lithography or etching equipment in cleanrooms. Let us talk about this development in more detail.
All modern HDDs use platters featuring a granular magnetic coating. In such media, grains vary slightly in size, shape, and exact positioning, which introduces noise and interference and limits how tightly bits (and tracks) can be packed. To read tracks reliably, HDD makers use specially designed two-dimensional (2D) read heads, which consist of two read heads. Disks used in Western Digital’s 10-platter 24TB HDDs with conventional magnetic recording and ePMR 2 feature an areal density of around 1.2 Tb/inch^2. Seagate’s 10-platter 30TB hard drive, which uses HAMR and FePt media, has an areal density of around 1.5 Tb/inch^2. Seagate believes that granular FePt (or other high-anisotropy media) will enable the industry to achieve areal densities of 4 Tb/inch^2 or possibly even 6 Tb/inch^2.
At around 5 Tb/inch^2, it will become harder not only to read granular media reliably but also to write to it, which will push the industry to move on to ordered-granular (OG) media that incorporates patterning in one dimension (according to Seagate). OG media is expected to feature grains of uniform size, placed at precise, repeatable intervals. This allows the write head to form data bits with tighter boundaries and less noise. OG media will require new magnetic coatings, but its benefits will be significant—with uniform grains, magnetic properties become more predictable, reducing variations in write/read performance.
However, ordered-granular media is expected to support areal densities of up to approximately 7 Tb/inch^2. To reach 8 Tb/inch^2, bit-patterned media—which physically carves the disk into isolated bits using lithography or other processes—will be needed. BPM represents a more radical shift in manufacturing and entails significant production costs.
Combining the physical isolation of bits (from BPM) with energy-assisted writing (from HAMR) in HDMR can push areal densities far beyond what either approach alone can achieve. This may explain why Seagate does not reveal its areal density expectations for this technology. However, BPM requires a complex patterning process (e.g., nanoimprint lithography, e-beam lithography, or highly sophisticated etching), which is expensive. Additionally, local heating must be extremely precise — too much heat affects adjacent dots, while too little results in write failures — so new types of lasers will need to be developed.
Given all the complexities associated with HDMR, this technology will not come to the market anytime soon. Western Digital expects it to arrive sometime in the next decade, at which point it will enable HDDs with capacities well over 100TB. Interestingly, the latest iteration of Western Digital's roadmap no longer mentions ordered-granular media, unlike its 2022 version. It is possible that this time, the company intends to move directly to the technology that delivers the best results.
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Anton Shilov is a contributing writer at Tom’s Hardware. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.


















-
maxx42 80TB, 100TB...For home users/consumers, that's a lot of data to lose if a drive fails. Or a lot of data to mirror, checksum, etc. to back up, which will take a LONG timeReply -
SomeoneElse23 maxx42 said:80TB, 100TB...For home users/consumers, that's a lot of data to lose if a drive fails. Or a lot of data to mirror, checksum, etc. to back up, which will take a LONG time
True this. So... that means we don't use large drives if we need the space? And if you don't need the space, clearly you don't go there.
Three in unraid including 2 parity drives would be relatively safe.
Last I checked, all data is only getting bigger. Not smaller. -
hotaru251 concern is how large can we actually go on HDD as would you be able to rebuild it before the time limit is up?Reply -
truerock
My old PC had: 1TB SSD, 12TB HDD, 6TB HDDmaxx42 said:80TB, 100TB...For home users/consumers, that's a lot of data to lose if a drive fails. Or a lot of data to mirror, checksum, etc. to back up, which will take a LONG time
My new PC that I built this year: 4TB SSD, 8TB SSD, 8TB HDD, 8TB HDD
It was very difficult to do full backups of the 12TB HDD. I'll probably never buy a HDD larger than 8TB in the future.
Also, it was expensive - but, I did purchase that 8TB SSD. For a home PC - it will be great when I can stop using HDDs and use only SSDs
About 12 years ago, I was a product manager for data backup for a large multi-national corporation.
Back in those days, the tape had to be re-copied every 2 years to make sure it was still readable.
The project I managed was converting all the tape backup to HDD backup.
I'm guessing all of the tape storage probably no longer exists in 2025.
I'm guessing if I still had that job today, we would be converting a lot of the data-backup infrastructure to SSDs. But, even today - there are applications for 100TB HDDs -
markhahn
Afaik home users haven't been a big part of the HD market for some years now.maxx42 said:80TB, 100TB...For home users/consumers, that's a lot of data to lose if a drive fails. Or a lot of data to mirror, checksum, etc. to back up, which will take a LONG time
At least not a direct part: consumer cloud storage is still spinning rust (indeed, sometimes sliding rust - tape).
The really interesting point here is the hat most data is quite cold: sometimes write-once-read-never, or at least not much is actually hot. And the fairly compact hot/warm stuff does just great on flash (which will always be several times more expensive by capacity). -
markhahn
Think of it more like tape, and recognize that these drives will almost always be part of a bigger system which manages resilvering, and which also contains lower-latency components.hotaru251 said:concern is how large can we actually go on HDD as would you be able to rebuild it before the time limit is up? -
lmcnabney It doesn't matter because the price per TB won't go down. HDD will be priced like GPUs.Reply -
King_V If you thought disk defragmenting took a long time before...Reply
... and I'll show myself out now. -
pug_s I am skeptical of this. 10 years ago, HD makers can only do 2TB per platter, and now it is 3. Hard drive size got bigger because they put more platters and stuffed Helium in the drive.Reply -
spongiemaster
Building an Unraid array or running a parity check with a 100TB parity drive would take over a week. That's not very practical.SomeoneElse23 said:Three in unraid including 2 parity drives would be relatively safe.