HAMR Reality Check - Do 30TB+ Drives Change Anything?

Executive Summary

Heat-assisted magnetic recording has arrived as a product rather than a roadmap item. Drives at 30TB and 32TB are shipping and we track them, which lets us answer the question that matters to a buyer rather than the one that interests the industry.

The answer is unflattering for the frontier. HAMR-class capacity currently carries a substantial premium per terabyte over the mainstream sweet spot - the frontier costs meaningfully more per terabyte than drives two-thirds its size. If your objective is capacity per pound, the newest drives are not where it is.

The second finding is about availability rather than price, and it is the one that could not be reported earlier. The frontier is thin. We track only a handful of HAMR-class rows and not all of them hold stock. 'Shipping' and 'buyable' are different claims, and the table on this page distinguishes them by showing out-of-stock rows rather than hiding them.

The truth table

The table below lists every frontier-capacity row we track, in stock or not. That inclusion is deliberate: a table that only showed buyable drives would answer 'what can I buy' while quietly failing to answer 'is this shipping in volume', and the second question is the one people actually have about a new recording technology.

We extend the table below the strict HAMR-class threshold to give a comparison band, and the recording technology is shown per row so the frontier drives are identifiable rather than blended in. Where a row is out of stock, it stays visible with its status.

The pattern to read: how many rows, how many buyable, and how the per-terabyte figures compare with the mainstream band underneath.

The premium, quantified

The stat block computes the cheapest live price per terabyte for the frontier class and for the mainstream 16-24TB band, and divides. The frontier is currently the more expensive way to buy a terabyte by a wide margin.

This is normal for a new areal-density generation and it is worth saying why rather than treating it as a scandal. Early production of a new recording technology is expensive, yields improve over time, and the first buyers of any density generation have always paid for the privilege. The premium is not evidence that HAMR is failing; it is evidence that it is new.

What it does mean is that the argument for buying the frontier is not economic. Anybody choosing 30TB drives today over 20TB drives is buying something other than value per terabyte, and it is worth being clear-eyed about what.

What HAMR actually does, and what it does not

HAMR increases areal density - how much data fits on a platter - by using heat to allow more stable magnetic media to be written. That is a genuine engineering achievement and it is the reason drive capacities can keep climbing at all.

What it does not do is relieve the current shortage, and conflating those two is the most common error in coverage of this technology.

The shortage is about drives per quarter, not terabytes per drive. The pressure documented in our sourcing - hyperscaler absorption reported by TrendForce in September 2025, enterprise orders queuing beyond a year per Tom's Hardware citing DigiTimes in November 2025 - is a manufacturing throughput and allocation constraint. A drive holding 32TB instead of 20TB does not make the factory produce more drives; it changes what each one holds.

It helps at the margin for buyers who are bay-constrained rather than budget-constrained, because fewer drives means fewer purchases from a constrained supply. But the notion that a density generation ends a supply shortage does not survive contact with the arithmetic.

Availability honesty

Which frontier models have actually held stock, as opposed to appearing and vanishing, is a question our tracking can answer within its window.

The frontier is thin. That is the finding. A category with a handful of tracked rows behaves differently from one with two hundred: a single seller going out of stock moves the whole category, and a price floor can jump because one listing disappeared rather than because the market moved. We would rather state that than present a small sample with the confidence appropriate to a large one.

Our seeding work on the Seagate lines recorded single-drive availability at the 28TB and above points that flickered rather than held - present one week, absent the next. Read the in-stock column on the table above as a current state rather than as a reliable prediction of what you will find tomorrow.

Why density generations always start expensive

The premium on this page is normal and it is worth understanding rather than being annoyed by, because understanding it tells you when to buy.

A new areal-density generation starts with low yields. The proportion of platters that meet specification is lower early in a process than late, and the drives that do meet specification carry the cost of the ones that did not. That cost declines as the process matures, and it declines fairly predictably - this is the same curve every storage density transition has followed.

There is also a demand effect working in the same direction. The first buyers of a new capacity point are, by definition, the buyers for whom capacity per drive matters more than cost per terabyte - hyperscale operators filling fixed rack space, and enterprises with bay constraints. Those buyers are not price-sensitive in the way a homelab buyer is, so early pricing is set against a market that will pay it.

Both effects unwind. The practical reading is that the frontier premium is a timing question rather than a permanent property of the technology, and that a buyer whose objective is value per terabyte should expect to wait rather than expect to be disappointed.

Rebuild time, and the argument against very large drives

There is a genuine technical argument against frontier capacities that has nothing to do with price, and it deserves stating alongside the economics because it does not go away when the premium does.

A larger drive takes longer to rebuild. When a drive fails in a parity array, the array reconstructs its contents onto a replacement, and the time that takes scales with the capacity of the drive. During that window the array is running with reduced or no redundancy, and it is working hard - which is precisely the condition under which a second drive is most likely to fail.

That is not an argument against large drives so much as an argument for dual parity as the array grows. Single parity on a twelve-bay array of 30TB drives is a configuration where a second failure during a multi-day rebuild is a real possibility rather than a theoretical one.

It also interacts with the recording-technology question in a way worth being explicit about. A shingled drive rebuilds far more slowly than a conventional one because the rebuild is a sustained random-write workload, which is the pattern shingled recording handles worst. Large plus shingled is the combination to avoid outright, and it is why every drive we list carries its recording technology.

Who should buy the frontier now

Two buyers, and the first is the real one.

The bay-constrained buyer. If you have a fixed number of drive bays and you need more capacity than the mainstream can put in them, the premium per terabyte is simply the price of the capacity existing at all. A twelve-bay chassis filled with 30TB drives holds substantially more than the same chassis filled with 20TB drives, and no amount of value-shopping changes the bay count. For this buyer the frontier is not a luxury, it is the only route.

The consolidation buyer, with caution. Fewer, larger drives means lower power draw, fewer things to fail, and fewer bays occupied. Those are real benefits. The caution is rebuild time: a larger drive takes longer to resilver, and a large array of very large drives spends a long time vulnerable after a failure. That argues for dual parity rather than against large drives, but it should be a deliberate decision.

Everybody whose objective is capacity per pound should buy the mainstream band and check back in a year. That is not a criticism of the technology; it is what the current numbers say, and the numbers on this page update themselves.

How to read a capacity record announcement

Frontier drives generate headlines out of proportion to their effect on anybody's purchase, and it is worth having a filter for them.

A capacity record is a statement about what one drive can hold. It is not a statement about how many drives exist, which is the number that determines whether you can buy anything. Those two get conflated constantly in coverage, and the conflation always runs the same direction: a density announcement gets reported as relief for a supply shortage that it does not address.

The questions worth asking when a new capacity point is announced are narrower and duller. Is it shipping in volume or sampling to qualified customers? At what price per terabyte relative to the mainstream band? And is it appearing in retail channels at all, or only through OEM allocation - which, given everything else in this report series, is where new production has been going first.

Our table answers the third question for the parts we track, and it answers it by showing out-of-stock rows rather than hiding them. A drive that is announced, reviewed, and not purchasable is a real state of affairs and a table that omitted it would be flattering the technology rather than describing the market.

The useful posture is neither excitement nor cynicism. Density transitions are how storage keeps getting cheaper over the long run, and they are reliably a bad deal in their first year. Both of those are true at once.

What to watch

The premium compresses as yields improve and as competing HAMR products reach the market. That is the normal path for a density generation, and the chart and table on this page will show it happening without anyone needing to write another report.

The specific thing worth watching is the crossover - the point at which the frontier's price per terabyte meets the mainstream band. When that happens, the buying advice inverts and large drives become the default rather than the specialist choice.

What will not be a signal, however it is reported, is a capacity record. A larger drive announced is not more terabytes shipped, and the constraint that matters for availability is drives per quarter.