Seagate: The HDD Shortage Is a Supply-Discipline Story, Not a Demand Story

The headline is convenient and mostly true: AI and cloud data centers are buying nearline hard drives as fast as Seagate and Western Digital can make them, and the industry's own math says supply will trail demand into 2028. That demand story is what has carried Seagate's stock up roughly 200% so far in 2026, to a market value near $190 billion. It is also the wrong story to build an investment view on.

The re-rating has far less to do with how much storage customers want than with how little the two remaining suppliers choose to make. Seagate can raise its exabyte output about a quarter each year while holding the number of drives it actually builds nearly flat, spending little on new capacity to do it. That is the entire economics of this cycle, and it is a supply phenomenon wearing demand's clothes.

The demand story is real, and it isn't the driver

Start with the demand. It is no fabrication. Morgan Stanley's research, published in June, puts the shortage in figures: nearline HDD supply runs roughly 300 exabytes short of demand in 2026, a 10–15% shortfall, widening toward a 400 exabyte gap in 2027 and 2028. The bank sees demand growing 40–50% a year against a supply base that can only add 30–35% annually. The evidence of tension is in the plumbing: hyperscalers are deploying drives at near 100% utilization against a historical norm around 70%, and distributor inventories have been squeezed to one or two weeks.

But here is the distinction that matters. Demand does not set the price when the suppliers can grow capacity faster than they choose to. An exabyte is an exabyte; the manufacturing cost of one is not fixed by how badly anyone wants it. What sets the price is the rate at which Seagate and Western Digital expand output — and that rate is deliberately slow.

The tell is in the capex line

There is no cycle in the storage business that a supply analyst reads faster than the capital-expenditure line. Seagate generated roughly $3.7 billion of operating cash flow over the trailing year while spending only about $570 million on capital expenditures. It produced about $3.1 billion of free cash flow. This is a company growing revenue 34% year over year with a gross margin near 41.5% and an operating margin around 28% — and reinvesting a fraction of its cash into new physical capacity.

That is not the behavior of a business chasing demand with new factories. It is the behavior of a business converting its existing footprint into pricing power. Seagate does not need more drives, and it is not building them. It needs more terabytes per drive, which it gets from a technology it spent a decade developing and can now turn on without standing up new wafer lines.

That technology is Heat-Assisted Magnetic Recording, or HAMR — essentially a laser that allows more data to be packed onto the same disk surface. The payoff is structural: as drives move from three to four terabytes per disk, Seagate grows its exabyte capacity while roughly holding unit output steady. By the end of its fiscal 2026, HAMR products accounted for about 40% of its nearline exabyte run-rate, with the second-generation platform ramping at the two largest cloud providers and a third-generation product on track for late 2027.

The consequence is where the margin lives. When a supplier can raise output largely by improving areal density rather than adding drives and buildings, incremental price increases drop through to profit almost untouched. Morgan Stanley's shortage math implies as much: blended nearline drive pricing sits near $14.30 to $14.90 per terabyte, supplier internal targets have moved to $25–30 per terabyte, and distributors without contracts are already selling in the $30–35 range. Rising price per terabyte, flat unit counts, near-zero added capex — that combination is supply discipline, not demand traction.

The constraint sits in two pairs of hands

Part of why discipline holds is that the market has become an oligopoly guarded by a wall. HAMR is not easy to copy: it requires years of certification and vertical integration across magnetics, materials science, and laser manufacturing, which is why the two incumbents treat it as a moat. Western Digital, the only other meaningful supplier, is itself running out of capacity; its own response is to chase Seagate with higher-capacity drives it is still developing.

So the bottleneck in storage has not moved to demand or to flash, which remains far more expensive per terabyte for bulk data. The constraint sits with the two firms that control how fast nearline capacity comes online — and both benefit from keeping it scarce. This is why the shortage is an earnings event, not a customer-service problem. Seagate cannot sell output it chooses not to build.

What is already in the price

The hard part is that the market understands the mechanism well enough to have paid for much of it. Seagate trades near 59 times trailing earnings and around 44 times trailing EBITDA, on a stock that has already tripled in a year. The forward case requires the discipline to hold for years, and expectations have moved to match: in its aggressive scenario Morgan Stanley sees Seagate's earnings per share roughly doubling again by fiscal 2028, versus consensus — and even its base case runs well above the Street.

Nothing in the shortage math forbids that outcome. But the same mechanism that built the rally can unwind it. If Western Digital brings meaningful capacity back, if hyperscalers stop signing long-term agreements, or if HAMR ramps faster than planned and Seagate must ship more units to defend share, the near-perfect flow-through to profit reverses just as quickly. The singular tell is the unit count. While exabytes rise and the number of drives shipped stays flat, pricing power holds; the moment Seagate starts adding units to chase share, the cycle reverts to the commodity that HDDs were before the industry finally stopped building.

This is not a question of whether AI demand persists. It will. The question is whether the two suppliers hold the line on capacity — and whether the drive count stays flat is how an investor watches it. Demand lit the fire; supply discipline is the engine that turned it into a 200% stock. Watch the units, not the exabytes.