Western Digital Corporation WDC

Key data

The business

Western Digital makes hard disk drives, the spinning-platter kind, and after separating its flash-memory arm, SanDisk, in early 2025 it does essentially only that. Nearly 90% of revenue is now "nearline" drives sold into cloud data centers, the high-capacity units that hold the warm and cold data behind video libraries, backups, and AI training sets. Two companies are left in this business worldwide after decades of consolidation buried everyone else: Western Digital and Seagate. The drive itself is a commodity sold by the terabyte, but the moat is that duopoly plus the capital and manufacturing yield needed to keep raising areal density, which is what keeps new entrants out. What moves the money is simple: how many exabytes ship, at what price, and whether the two makers stay disciplined instead of flooding the channel the way they did in 2022 and 2023.

The numbers

The story here is a business coming off the floor at a rate that is easy to underappreciate. Revenue and drive profitability have climbed every single quarter for a year and a half.

Gross margin went from 41% to 54% in five quarters, an unusually clean ramp for a commodity maker and the tell that pricing, not just volume, is doing the work. The EPS line flatters the picture: the jump to $8 in the back half carries a large paper gain on the retained SanDisk stake, so drive EPS is a fraction of it. Non-drive earnings ran roughly $1.78 to $3.56 per quarter on the basis analysts track, and the company cleared their number all four quarters (Q4 was $3.56 against $3.31 expected).

FY22 included flash and is not comparable; the honest baseline is FY24's $6.3B and a loss. From there revenue doubled and margin nearly doubled in two years. The distortion flagged two days ago still holds: the 10-K confirms about $5.5B of last year's $9.9B pretax profit came from outside the drive business, so the $24.28 GAAP figure overstates what the drives earn. On the Street's non-GAAP basis WD earned about $10.19 in FY26. Consensus then models $32.54 in fiscal 2028 and $45.03 in fiscal 2029, a near-tripling. At $467 you pay about 14 times the FY28 number and 10 times FY29. The variant question is narrow and everything rides on it: is a 54% gross margin the new floor of a rational duopoly, or a cyclical top that mean-reverts the way FY23 did. The print that settles it is exabytes shipped and price per exabyte, disclosed annually, plus whether margin holds above the mid-40s through the next two quarters.

The lens for a drive maker is who buys it, and the answer is concentrating: three cloud buyers are now 44% of sales, up from 39%. That is the engine and the fragility in one row.

Management

The capital story turned decisively this year. Long-term debt fell from $4.71B to $1.05B, cash-funded, and the company moved from token buybacks of $149M in FY25 to $2.59B in FY26 alongside a new dividend. That is the right instinct after years of leverage, though the buybacks were struck after the stock had already run from $92 toward $800, so the timing was rich, not cheap. Insiders sold about $13.2M over the past year and bought nothing; CEO Irving Tan sold about $6.1M on August 11, plan status not disclosed. Against a $161B market value that selling is noise, not a signal. Pay leans to the profit surge that is partly the SanDisk gain, which is worth watching when the compensation proxy lands.

How it fails or surprises you

Margin mean-reversion. Gross margin at 54.1% is an all-time high for a business that has cratered before. If cloud buyers digest their builds and orders pause, price rolls over first and margin can retrace toward the 30s, as it did in FY23 and FY24. The first warning is a quarter with margin under 45% and revenue down sequentially.

Customer concentration. Three hyperscalers are 44% of revenue and rising. Any one of them dual-sourcing more to Seagate, building its own storage tiers, or simply pausing a cycle punches an air pocket in a single quarter. Watch the largest-customer share and any sequential revenue drop that margin cannot explain.

Structural nearline demand holds (right tail). If AI-driven data growth keeps nearline exabytes compounding double digits and the two-maker structure keeps pricing firm, the FY29 estimate of $45 proves conservative and margins stay above 50%. The market still anchors WD to its boom-bust past. Double-digit exabyte growth with flat-to-higher price per exabyte would reveal it first.

Closing thoughts

Gross margin over the next two to three quarters, plus the annual exabyte and price-per-exabyte disclosures, will show whether a two-maker market keeps pricing firm or this is another commodity peak mistaken for a new era. The cloud companies' capital spending underneath won't resolve on anyone's timetable. The left tail is real and is precisely why the stock sits at 14 times forward rather than 25: reported earnings are inflated by a one-time gain, margins are at a record, and 44% of the business rides on three buyers. The right tail, a permanent re-rating as the market accepts that two disciplined makers is a structurally better business than the six that came before, is live but you are paying a fair price for it, not a bargain. If margins revert, EPS and the multiple compress together, which is how you lose half.

The bet is still that the big cloud companies keep buying more high-capacity hard drives to hold the flood of data their AI systems create, and that Western Digital, one of only two makers left, gets its share. What breaks it is the pair that always breaks it here, gross margin slipping under 45% while revenue rolls over in the same quarter. Hold the memo to that line: if the next two prints show margin above the mid-40s with exabytes still climbing, the duopoly thesis is intact; if not, this was the top of another cycle dressed up as a new era.