Everything About AI
NextFin News - Seagate Technology fell 11% and Western Digital dropped 7% in early Friday trading after a report that Toshiba plans to invest roughly ¥60 billion — about $400 million — to double its hard disk drive production capacity by fiscal 2027, the Japanese company's first meaningful hard-drive investment in roughly half a decade. The selloff strikes at the heart of the rally that has made the two storage makers among the market's best performers this year: a belief that a tight, disciplined three-supplier market would keep pricing power intact as AI data centers soak up every available exabyte.
The timing is the sting. Western Digital has said its long-term hard-drive capacity is fully booked through calendar 2026, with hyperscale customers locking in multi-year supply. Seagate's chair and chief executive, Dave Mosley, said on the company's July earnings call that the vast majority of nearline exabytes are already allocated. Toshiba's re-entry — with the new capacity targeted for fiscal 2027, the Japanese company's year ending in March — is a reminder that no capacity shortage survives high margins untouched.
The Report: A Third Player Re-Enters a Two-Horse Race
According to a report in Japan, Toshiba aims to double production of hard disk drives used in artificial-intelligence data centers, centering the expansion on its factory in the Philippines. The company currently accounts for just over 10% of the HDD market by capacity but has set a medium-term target of reaching 30%. The investment, about ¥60 billion ($379.4 million), marks Toshiba's return to serious hard-drive capital spending after years in which the industry consolidated around Seagate and Western Digital.
The structure of the market is what made the rally work. Three companies — Seagate, Western Digital and Toshiba — dominate global hard-disk supply, and the two U.S. rivals have spent the past cycle behaving less like commodity vendors and more like a disciplined duopoly. Western Digital's nearline share reached 49% in the latest quarter, the highest since 2019, while Seagate's stood at 39%, its lowest in the same span, according to Barclays analyst Tom O'Malley. That nearline segment is the profit engine: exabyte shipments of high-capacity nearline drives are forecast to grow 14% this year, to 66.6 million units, before flattening in 2026.
Into that setup walks Toshiba with a stated ambition to triple its share. The market's read was immediate and unforgiving: if the smallest player can commit fresh capital, the capacity discipline that underpinned the pricing-power trade was never as binding as investors assumed.
Why the Selloff Is About 2026 Profits, Not Just 2027 Capacity
The obvious question is also the wrong one. Toshiba's new capacity lands in fiscal 2027 — months, not years, away — while Western Digital's and Seagate's order books are full through calendar 2026. On a pure supply-and-demand timeline, the report should not move this quarter's earnings by a yen.
But equity markets do not discount the quarter that is booked; they discount the quarter in which the booking power slips. The transmission channel here is the multiple, not the near-term revenue line. Investors have been willing to pay up for Seagate and Western Digital because they believed the industry had finally escaped its old cycle: build capacity, crash prices, lose money, stop building. A 10%-share player announcing a doubling of capacity is the first visible crack in that belief, and the market repriced the terminal value of the pricing-power thesis before a single new drive ships.
Technology transitions alone should be enough to grow EB [exabytes] at mid-20% CAGR.
That line, from Barclays' O'Malley, captures the oddity of the moment. The HDD industry is growing exabytes while shipping fewer units — a structural shift toward larger, denser, more expensive nearline drives that has been the bull case for both stocks. Seagate is up 124% year to date and Western Digital 111% ahead of Friday's drop, per the same analyst note. The risk is not that demand disappears. It is that the pricing premium required to fund that exabyte growth evaporates once a third competitor decides the margins look attractive again.
This is the second-order move that explains the size of the drop. The first-order effect — more supply in 2027 — is small and distant. The second-order effect is a repricing of the entire investment case: if the industry is not a disciplined oligopoly, it is a capital-intensive commodity business, and commodity businesses do not command the multiples that pricing-power stories do. The market moved first on the valuation regime, and only second on the supply math.
Cyclical Shock Inside a Structural Uptrend
Is this the end of the storage rally, or a buying opportunity? The answer requires separating two forces that the headline conflates. The demand side is structural; the supply side is cyclical. Getting this backwards flips the conclusion.
The demand side is structural because the economics of AI storage are not changing. AI data centers do not store training sets, inference logs and video archives on flash — the cost per terabyte still favors hard drives at scale, and that arithmetic is not changing. Global HDD revenue is projected to reach $22 billion this year, up 22%, and $23.4 billion in 2026, according to Barclays. Exabyte shipments surpassed 450 exabytes by late 2025 and quarterly industry revenue has held above $6 billion into 2026. This is a market whose demand curve has shifted up, not one in terminal decline.
The supply side is cyclical, and the history is unambiguous. Every margin recovery in the hard-drive industry has eventually summoned new supply, and every supply wave has ended in a price reset. Three cycles prove the pattern. After the 2011 Thailand floods wiped out roughly half of global hard-drive capacity, prices spiked and every surviving producer raced to rebuild — only to face a multi-year price decline once capacity returned. In the 2016-2017 upcycle, strong data-center demand lifted pricing until new capacity arrived and the industry entered the consolidation that ultimately pushed out smaller players. And in the 2020-2021 pandemic demand surge, the same sequence repeated: demand jumped, capacity followed, and by 2023 the industry was in a downturn that reset the baseline.
Toshiba's ¥60 billion is small next to the tens of billions Seagate and Western Digital have spent on HAMR, ePMR and UltraSMR roadmaps, but it is large enough to signal that the discipline is negotiable. Cyclical supply shocks revert: once capacity lands and average selling prices compress, marginal producers stop building again. The fiscal 2027 capacity addition is a mean-reverting event inside a structurally higher demand path.
That distinction matters because it says the selloff is rational at the margin but probably excessive at the core. The pricing-power premium embedded in these stocks is at genuine risk — that is the cyclical leg. The exabyte growth thesis is not — that is the structural leg. Investors who treat the two as one will either sell a structural compounder on a cyclical headline or buy a commodity multiple back to life.
The Asymmetry: Western Digital Has More to Lose Than Seagate
Not all of the pain is shared equally. Western Digital entered this episode with more to protect. Its 49% nearline share — the highest since 2019 — was the crown jewel of the bull case, and its capacity is fully booked through 2026 on multi-year contracts. A market that moves toward Toshiba's 30% aspiration erodes exactly the share premium Western Digital was trading at. Seagate, at a 39% nearline share, has less dominance to defend, and its stock had already begun to lag Western Digital's parabolic run before Friday.
The earnings backdrop underscores the asymmetry. Western Digital reported fiscal fourth-quarter revenue of $3.75 billion, up 44% year over year, with 231 exabytes shipped and non-GAAP gross margin of 54.4%. Seagate posted fiscal fourth-quarter revenue of $3.63 billion, up 48.5%, with HAMR-based Mozaic drives now accounting for 40% of nearline exabyte shipments and annual free cash flow of $3.1 billion. Both companies are printing money — which is precisely why Toshiba's re-entry stings. High returns advertise opportunity to anyone with a factory and a balance sheet.
There is also a technology asymmetry that cuts the other way. The next generation of nearline density — Western Digital's ePMR and UltraSMR, Seagate's HAMR roadmap — is where the margin sits. Toshiba's expansion is a capacity play on today's density frontier; if the leaders are already two generations ahead by the time that capacity arrives, the new supply competes at the low end of the curve while the leaders price the high end. That is the optimistic case for the incumbents, and it is why a hard-nosed investor would want to see nearline average selling prices before declaring the duopoly broken. Capacity can be built in quarters; a density roadmap takes years.
The Counter-Thesis: Discipline Was Always an Illusion
The strongest case against the "overreaction" read is simple and historical: hard-drive capacity discipline has broken exactly when it looked most secure. This is a three-player market where every margin upswing for two decades has ended in a supply response. Toshiba's move is not an anomaly; it is the predictable behavior of a rational third competitor watching the other two print money. In that light, Friday's drop is not panic — it is the market finally crediting the oldest rule in the industry.
There is weight to that view. A 10-point share swing in a concentrated market does not require Toshiba to win; it only requires Toshiba to be present. If hyperscale buyers — Amazon, Microsoft, Alphabet, Meta — see a credible third source, their willingness to accept the incumbents' pricing hardens. Multi-year contracts protect 2026; they do not protect the 2027 renewal. Once a buyer has a qualified second or third source, the renewal is no longer a negotiation between a captive customer and a monopolist — it is a competitive bid.
There is also a capital-market angle the bull case tends to skip. Seagate and Western Digital have spent the past two years buying back shares and retiring debt, returning cash to shareholders because they saw limited high-return internal projects. Toshiba's decision to deploy ¥60 billion into the same market is, in effect, a competing claim on where the best returns are. If a third player sees attractive returns at today's prices, the incumbents' own capital allocation — and the buyback-fueled EPS growth that supported their multiples — comes under scrutiny.
The falsifying signal is concrete: watch nearline average selling prices through fiscal 2027. If average selling prices hold even as Toshiba ramps, the incumbents' technology moat is real and Friday was a sentiment overshoot. If nearline ASPs compress by more than 10% while exabyte demand keeps growing, the pricing-power thesis is broken and the multiple has further to fall. A second tell: Toshiba's own execution. If its capacity ramp misses its stated target by a quarter or more, the threat was paper, not silicon.
What Comes Next
Short term, the stocks trade on sentiment and on the next round of capacity commentary from Seagate and Western Digital themselves — any hint that the leaders are accelerating their own capex to defend share would extend the multiple compression. The next earnings calls, due in late October, will be scanned for any change in tone on allocation, pricing and the durability of those 2026 bookings.
Medium term, the numbers that matter are nearline ASPs, exabyte shipment growth and the renewal terms on those 2026 contracts. A stable ASP print in the December quarter would do more to repair the shares than any commentary. Long term, the question is whether AI data-center demand can absorb three committed suppliers without resetting the industry's return profile.
The base case is a split verdict: a cyclical de-rating of the pricing-power premium, overlaid on a structural demand trend that keeps revenues growing. The upside case is that Toshiba's ramp stumbles and the incumbents' density lead keeps ASPs firm — in which case Friday becomes a footnote. The downside case is that capacity discipline was indeed an illusion, ASPs roll over in 2027, and the market discovers that a 30% aspiration from the smallest player is enough to reset the whole industry's multiple.
The market just learned that the hard-drive duopoly's most valuable asset was never its capacity — it was the belief that no one else would build any. Toshiba has now announced that it will. The pricing power that funded this rally was always rented, not owned, and the lease just came up for review.
Explore more exclusive insights at nextfin.ai.
