Seagate Has Two Engines — Industry Pricing And A Cost Curve Only It Controls

The story behind this stock

Seagate is the one position in my portfolio where the bottleneck has no substitute technology.

That is a stronger statement than it sounds. Memory shortages resolve when someone builds a fab. Optical shortages resolve when someone builds more laser capacity. Nearline hard drives are different: they are the only economically viable way to store data at hyperscale, the cost-per-terabyte gap versus flash is widening rather than narrowing, and the industry consolidated to three players who spent the last decade not adding capacity. There is no alternative waiting in the wings and no idle fab to restart.

The arithmetic of that shows up in the supply gap. Nearline HDD supply is expected to fall roughly 300 exabytes short of demand in 2026 — a 10–15% shortfall — widening to something like 400 exabytes in both 2027 and 2028. Blended nearline pricing has been running near $14.30–14.90 per terabyte against scenarios that put it meaningfully higher over the next two years.

The clearest evidence of who holds power in that negotiation isn't a price. It's contract duration. Cloud providers would like long-term agreements running out to 2032. Seagate and Western Digital are declining to commit commercial terms beyond twelve months. Sellers refuse to lock in prices when they expect prices to rise.

What actually changed — and why the June quarter was the important one

Seagate closed fiscal 2026 with a June quarter that beat on every line and then guided far above what anyone modelled.

Revenue of $3.63b came in 48% above the prior year and 17% sequentially, against a $3.49b consensus. Adjusted EPS of $5.71 landed $0.62 ahead of the $5.09 estimate. Non-GAAP gross margin reached 52.7%, expanding for a thirteenth consecutive quarter, with operating margin at 44.6%. Free cash flow hit $1.12b in the quarter and a record $3.1b for the fiscal year.

Then the guidance: first-quarter fiscal 2027 revenue of roughly $4.1b, up about 56% year over year, with non-GAAP EPS near $7.30 and gross margin approaching 57%. Management said orders provide visibility running toward 2029.

But the number I care most about is 40%. HAMR-based Mozaic drives closed the fiscal year at roughly 40% of nearline exabyte shipments, hitting the milestone management had targeted for exactly that quarter. Mozaic 4 — supporting up to 44 terabytes per drive — is ramping at major cloud customers, with 50% of HAMR exabytes expected on that platform by the end of calendar 2026 and 70% HAMR penetration of nearline exabytes targeted by June 2027.

CFO Gianluca Romano described incremental gross margin as well above 60%, and tied the margin trajectory directly to the HAMR ramp rather than to pricing.

That distinction is the whole investment case, and it's the next section.

Two engines, not one

This is what separates Seagate from the other storage positions I hold.

Engine one is the industry. Supply is short, pricing is rising, and Seagate captures it. Every incremental dollar of nearline price requires no additional cost and flows almost directly to the bottom line. This engine is cyclical and I do not control it.

Engine two is areal density, and it belongs to Seagate alone. HAMR increases the number of terabytes on each platter. More capacity per drive means fewer drives, fewer heads, fewer platters and less assembly labour per exabyte shipped. CEO Dave Mosley calls areal density the company's "North Star" and describes it as the fastest and most capital-efficient route to long-term exabyte growth — which is a precise way of saying it expands supply and lowers unit cost at the same time.

This matters enormously for what happens when the cycle turns. A pure pricing bet loses everything when ASPs normalise. Seagate's cost curve keeps falling regardless, because HAMR penetration is going from 40% toward 70% on a schedule set by qualification milestones rather than by the pricing environment. Thirteen consecutive quarters of gross margin expansion did not all happen in a shortage.

Western Digital is ramping its own HAMR programme, so this advantage is a lead rather than a moat. But leads in areal density have historically persisted for years, because the qualification cycle at each hyperscaler is long and customers are slow to re-qualify.

How this fits the rest of the book

I run eight positions at equal weight inside a single theme. Three of them — Micron, SanDisk and Seagate — are bets on the scarcity of stored bits, and I'd rather name that concentration than have a reader discover it.

They are genuinely different shortages. Micron is DRAM and HBM, constrained by wafer allocation to a product with three times the area intensity. SanDisk is NAND, where industry capex grew only about 5% year over year. Seagate is nearline HDD, where no new capacity has been built in a decade.

There is one specific correlation worth flagging: the HDD shortage is actively pushing hyperscalers to shift some storage to SSDs, which tightens NAND. So Seagate and SanDisk partially feed each other, meaning they are more correlated than a layer map suggests. Both benefit from the same underlying scarcity of stored bits.

What makes Seagate the most defensible of the three is engine two. Micron and SanDisk are close to pure ASP bets with contract structures layered on top. Seagate has a company-specific cost lever running on its own timetable. If I had to hold one storage position through a cycle turn, it would be this one.

My Lam Research holding is the partial hedge across all three, since the equipment that eventually relieves these shortages is what Lam sells.

Why Seagate rather than Western Digital

Both are pure HDD businesses post-spin and both benefit from the same duopoly shortage. I went with Seagate on three grounds.

Technology lead. Seagate is further along the HAMR transition, with 40% of nearline exabytes already shipping on Mozaic and a defined path to 70% by mid-2027. Western Digital's HAMR ramp is behind, which means Seagate's cost-per-terabyte advantage should widen before it narrows.

Operating leverage. Morgan Stanley's scenario work illustrates the gap: under an optimistic pricing path, Seagate's CY27 EPS could exceed $70 and CY28 could exceed $100, versus above $40 and $70 for Western Digital — against CY25 actuals of $10.24 and $6.95 respectively. Those are scenarios rather than forecasts, but the relative leverage is the point.

Balance sheet trajectory. Seagate reduced gross debt by $1.4b during fiscal 2026 and retired a further $1b of high-yield notes in July. Converting a cyclical peak into deleveraging rather than into capacity is exactly the discipline that determines whether the next downturn is survivable.

Catalysts I'm watching

HAMR penetration hitting 50% then 70%. These are dated, specific milestones. They are the cleanest read on engine two, and they are largely independent of the pricing environment.

Gross margin approaching 57% in the September quarter. Guidance is explicit. A print at or above that validates the incremental-margin comment; a miss would suggest mix or cost problems.

Contract duration. If Seagate starts agreeing to multi-year pricing, that tells me management thinks the peak is near. If it keeps refusing beyond twelve months, it doesn't.

Western Digital's HAMR qualification timeline. The single biggest threat to the margin story is the competitive gap closing faster than expected.

Capacity discipline. The industry has not added substantial capacity in a decade. If either player announces meaningful greenfield expansion, the entire duopoly premise is on a clock.

Assumptions behind my fair value

Revenue. First-quarter fiscal 2027 guidance of $4.1b implies a run rate above $16b. My model assumes continued growth as exabyte demand compounds and HAMR increases capacity per drive, but decelerating — 56% year-over-year growth is a shortage artifact, not a run rate.

Margins. Non-GAAP gross margin reached 52.7% with the September quarter guided toward 57% and operating margin near 50%. I'm modelling margins that stay well above historical norms but below the guided peak, because part of the current level is pricing that I expect to normalise even as HAMR costs keep falling. The net margin assumption is the one that carries the model.

Future P/E. I'm deliberately restrained here. Hard drives have always traded at a low multiple, and the argument that HAMR plus duopoly discipline earns a re-rating is exactly the argument every cyclical makes at the top. My exit multiple stays in the range HDDs have historically commanded. If the re-rating happens, it's upside I haven't paid for.

Share count. Roughly 210m diluted. Record free cash flow supports buybacks, but management has prioritised debt reduction and I've modelled the count flat rather than assuming shrinkage.

What would make me wrong

The duopoly breaks. Everything here depends on two suppliers staying disciplined about capacity. Nothing enforces that except each company's own judgement, and the incentive to defect grows with every quarter of 50%+ margins.

Western Digital closes the HAMR gap faster than expected. One of the bear cases on the street assumes gross margins peak in the low-to-mid 40s as Western Digital's HAMR ramp intensifies pricing competition through 2027. That is a credible scenario, not a strawman.

The cycle turns and the multiple compresses with the earnings. This is the classic cyclical trap and it cuts twice — peak earnings valued at a peak multiple falls much further than either component alone. Seagate rose roughly 445% in a year through July. There is no valuation cushion.

Substitution risk over a longer horizon. I've argued there's no alternative to nearline HDD. That's true today. It is less certain across five years if NAND cost curves fall faster than expected or a new archival medium reaches production scale.

AI capex financing. Shared across every position I hold. Roughly $700–725b of hyperscaler spending, increasingly debt-funded against operating cash flow estimates suggest is already ~94% committed.

Execution on qualification. HAMR is technically difficult. A qualification slip at a major cloud customer would delay the ramp that carries the margin assumption, and the stock prices in the schedule being met.

Where I land

Seagate is the storage-capacity position in a portfolio built on physical scarcity, and it's the one I'd defend furthest into a downturn.

Not because the shortage is more durable than the others — it may or may not be. Because Seagate is the only one of my three storage holdings with a margin lever that runs independently of the pricing cycle. HAMR penetration going from 40% to 70% happens on a qualification schedule, not a price schedule. That's a real asset in a portfolio otherwise dependent on shortages persisting.

My sell signal is specific and it isn't a price: either player announcing meaningful greenfield capacity expansion, or Seagate agreeing to multi-year fixed pricing with a major cloud customer. The first says the duopoly is ending. The second says management thinks the peak has arrived. Neither has happened, and the refusal to price beyond twelve months tells me a lot about what the people closest to this actually believe.