HAMR increases HDD areal density by briefly heating the recording media as data is written, allowing smaller magnetic regions to remain stable once they cool. As the technology moves into hyperscale production, those density gains are beginning to change the infrastructure economics of capacity storage.
For decades, HDD capacity has increased dramatically while the standard enterprise drive footprint has little changed. Although the familiar 3.5-inch enterprise form factor has remained broadly constant, what has changed is how much hard drive manufacturers can pack onto the platter surfaces inside it.
That makes areal density—the amount of data stored on a given area of magnetic media—one of the most important variables in the economics of HDD capacity. As datasets grow faster than physical infrastructure can expand, the industry increasingly depends on each drive carrying more data rather than simply deploying proportionally more drives.
Why Areal Density Matters in HDDs
Areal density determines how much information can be written onto each platter surface. Raise it, and the effect compounds through the drive: more bits per square inch become more terabytes per platter, and more terabytes per platter become more capacity within the same basic 3.5-inch form factor.
For data centers, that matters well beyond the drive specification. If each HDD stores more data, a given capacity target can potentially be met with fewer drives, fewer enclosures, and less supporting infrastructure—or an existing footprint can absorb more capacity without growing at the same rate. Power and cooling do not fall in perfect proportion to drive count, but higher density can improve the amount of storage delivered per rack and per watt.
The market is already moving in that direction. TRENDFOCUS forecasts cited by SNIA project nearline HDD unit shipments to grow at a much slower rate than shipped capacity, with unit CAGR around 6% compared with capacity CAGR above 24%. The implication is simple: the industry expects substantially more data to be stored without anything like the same increase in the number of physical drives.
The Magnetic Recording Problem HAMR Solves
At its simplest, magnetic recording depends on setting tiny regions of a platter into one of two stable magnetic states and then reading those states back reliably. Most modern HDDs use perpendicular magnetic recording (PMR), in which the magnetization is oriented perpendicular to the platter surface.
The underlying problem is easier to picture with the light-switch analogy. A recorded bit needs to behave like a switch that points decisively in one direction or the other. It has to be easy enough for the write head to flip when new data is written, but stable enough to stay where it was put for years afterward.
To fit more data onto the same platter surface, manufacturers must make the recorded regions—and the magnetic grains that support them—smaller. At this reduced scale, the properties that keep data stable begin to work against the ability to write to the media.
Stability, Writability, and Smaller Bits
The challenge is real. Smaller magnetic grains are more vulnerable to thermal fluctuations, which can disturb their magnetic state over time. To keep recorded data stable, manufacturers therefore need media with greater magnetic anisotropy—the tendency of the material to hold its magnetization in a preferred direction.
That increased stability brings its own problems: the more magnetically “hard” the media becomes, the stronger the field required to change its state. As researchers at the University of York explain, the write field available from an HDD head is ultimately limited by the magnetic materials used in the head itself. Density, thermal stability, and writability therefore begin to pull in different directions.
This is the constraint HAMR is designed to break. The storage medium needs to remain difficult to disturb during normal operation, but become temporarily easier to write at the exact moment new data is recorded.
How HAMR Uses Heat to Increase Recording Density
HAMR changes this tradeoff by heating only the area being written, and only for an instant. Raising the temperature temporarily reduces the magnetic hardness of the recording media, allowing the write head to change the state of material that would otherwise be too resistant to write reliably. Once the spot cools, the media regains its high stability and the newly written bit remains fixed.
In Seagate’s current HAMR implementation, a small laser diode on each recording head sends light through an optical waveguide to a near-field transducer, which concentrates that energy onto a microscopic area of the platter immediately as data is written. The heating, writing, and cooling cycle takes place within about a nanosecond.
That timing matters. HAMR does not work by broadly heating the platter or the drive. It creates an extremely localized thermal assist at the moment of writing, allowing manufacturers to use media engineered for stability at normal operating temperatures without requiring the magnetic write head alone to overcome that same resistance.
Achieving that reliably is an engineering problem of its own. The laser, optical components, recording head, media stack, and heat flow all have to work together without allowing the thermal pulse to disturb neighboring data.
Those constraints helped turn HAMR into a decades-long development project. They also make its arrival into hyperscale production significant: the technology has moved from a laboratory experiment into a manufacturable solution.
HAMR Has Moved From Roadmap to Production
For years, HAMR was a technology more readily featured on manufacturer roadmaps than in deployed storage fleets. That is no longer the case. In March 2026, Seagate said its Mozaic 4+ platform had been qualified and put into production with two leading hyperscale cloud providers, with HAMR drives supporting capacities up to 44TB shipping in volume.
Mozaic 4+ exceeds 4TB per disk while retaining a 10-disk architecture. The capacity gain is therefore coming primarily from putting more data onto each recording surface rather than adding more platters inside the drive. Seagate says broader availability will follow as production continues to expand.
However, while Seagate has crossed the production threshold with HAMR, Western Digital and Toshiba are still pursuing different combinations of recording technologies and capacity gains. Not every high-capacity HDD is suddenly relying on HAMR.
HAMR, ePMR, MAMR, CMR, and SMR: How They Relate
The alphabet soup around modern HDD recording can make these technologies sound like competing alternatives. They are not. The clearest way to understand them is to separate two different questions: how the bits are written, and how the tracks are arranged on the platter.
WD makes the distinction explicit. PMR describes the perpendicular recording structure that underpins modern HDDs. CMR and SMR describe recording formats: CMR keeps tracks conventionally separated, while shingled magnetic recording (SMR) overlaps neighboring tracks to increase track density. SMR can therefore add capacity, but it also changes write behavior and requires different workload and host-management considerations.
By contrast, ePMR, MAMR and HAMR are energy-assisted approaches used to make increasingly dense magnetic media writable. WD’s ePMR applies additional electrical current to the write head to strengthen the write process. With MAMR, microwave-assisted magnetic recording uses microwave energy to help switch the magnetic state of the media.
A drive can use an energy-assisted recording technology while also using either a conventional or shingled track layout. Toshiba’s current M12 is a useful example: the company combines FC-MAMR with host-managed SMR to reach 30–34TB. WD is likewise qualifying a 40TB UltraSMR drive based on ePMR while separately qualifying HAMR with hyperscale customers.
That overlap is important when reading capacity roadmaps. HAMR does not arrive as a clean replacement for every technology that came before it. Manufacturers can continue extracting gains from platter counts, track layout, media design, head technology and existing recording platforms while introducing HAMR where its additional areal-density potential justifies the engineering and qualification effort.
What Higher Areal Density Changes in the Data Center
The practical value of higher areal density is that it changes how much infrastructure is required to deliver a given amount of raw capacity. More terabytes per platter become more terabytes per drive; more terabytes per drive can become more capacity per enclosure and per rack. For operators working within fixed floor-space, rack and power budgets, that can be as important as the headline drive capacity itself.
The effect is not perfectly linear. A higher-capacity HDD does not eliminate every watt of power or every piece of supporting infrastructure that would have been attached to a lower-capacity drive. Controllers, networking, redundancy, cooling and operational overhead remain. But when capacity increases faster than per-drive resource consumption, metrics such as watts per terabyte and terabytes per rack can improve. Seagate makes that case directly for Mozaic, arguing that its higher areal density raises storage efficiency per square foot while reducing watts per terabyte. Those are vendor claims, but the underlying relationship is straightforward: storing more data in each drive slot changes the infrastructure denominator.
Industry forecasts point in the same direction. TRENDFOCUS data cited by SNIA projects nearline HDD unit shipments growing at about 6% CAGR through 2029 while shipped capacity grows at more than 24% CAGR. The expected growth is therefore not primarily a story of data centers installing proportionally more drives. It is a story of each generation carrying substantially more data.
That matters for the economics of the capacity tier. HDDs remain attractive where workloads can tolerate their latency and performance characteristics because they deliver large amounts of persistent storage at low cost per terabyte. Higher areal density strengthens that role by allowing operators to increase capacity without expanding the physical storage footprint at the same rate. It can also change the point at which a denser generation becomes worth qualifying or introducing into a fleet.
Density alone, however, is not a refresh strategy. A higher-capacity drive still has to fit the workload, platform, qualification process, failure model and procurement economics of the environment around it. The value of HAMR is therefore not simply that it creates a larger HDD. It extends one of the mechanisms by which the capacity layer can keep scaling without requiring the rest of the data center to grow in direct proportion.
How Far Can HAMR Scale?
The useful question is not whether HAMR can support one more step up in drive capacity, but how much additional areal-density headroom it can create over multiple generations. Seagate’s current Mozaic 4+ platform has moved beyond 4TB per disk, and the company says its HAMR roadmap extends toward 10TB per disk—enough, in principle, to support drives of up to 100TB within the same basic 3.5-inch form factor.
Western Digital is taking a more gradual route. Its 40TB UltraSMR ePMR drive is in hyperscale qualification, and the company says ePMR can continue toward 60TB while HAMR is qualified in parallel. WD’s stated roadmap calls for HAMR to scale to 100TB by 2029, with production ramp targeted for 2027. The overlap is deliberate: customers can continue qualifying higher-capacity ePMR generations while HAMR matures into the next density step.
Toshiba’s path is different again. The company is extending MAMR alongside higher platter counts, including a verified 12-disk architecture that it plans to combine with MAMR for 40TB-class drives in 2027. At the same time, Toshiba says it is developing both MAS-MAMR and HAMR as next-generation recording technologies. Its roadmap is a reminder that future HDD capacity will continue to come from combinations of recording physics, media, track layout and mechanical design rather than from HAMR alone.
These are manufacturer roadmaps, not guaranteed delivery schedules, and the dates and capacity points will continue to move. The more durable takeaway is that HAMR gives the industry substantially more room to keep increasing areal density after earlier recording approaches begin to flatten. A 100TB-class drive is an attention-grabbing milestone; the more important development is the possibility of continuing to raise capacity per platter, per drive and per rack across several product generations.
The Future of HDD Capacity Is a Density Problem
The long-term challenge for HDDs is not simply to make each new generation larger. It is to keep increasing capacity without requiring the physical storage footprint, power envelope and supporting infrastructure to grow at the same pace as the data being stored.
That is why areal density matters more than any single headline capacity. Adding platters, using SMR and refining existing recording technologies can continue to extend drive capacity, but each approach has practical limits. HAMR opens another path by allowing the magnetic media itself to support smaller, stable recording regions and therefore more data on the same platter surface.
As that capability moves into production, the significance of HAMR becomes less about proving the technology and more about what higher density makes possible. More terabytes per platter can become more terabytes per drive and per rack, changing the economics of large-scale capacity storage even before the industry reaches the 100TB-class drives on current roadmaps.
The next era of HDD capacity will still depend on workload fit, qualification, reliability and cost. But the underlying engineering problem is increasingly clear: storing more data at scale means extracting more useful capacity from every square inch of magnetic media. HAMR matters because it gives manufacturers more room to keep doing exactly that.
Get in touch with Horizon to discuss how you can manage the lifecycles of your storage hardware.
