The Storage and Memory Signal

Direction: Worsening.

This period: Western Digital's fiscal Q1 2027 results, reported October 22, put a number on just how little room this market has: revenue up 55% year over year to $4.38 billion, average price per terabyte up 28%, and management describing the nearline hard drive market as a "25%-plus CAGR" business inside a "very tight supply environment," with guidance for 66 to 72% revenue growth next quarter that only makes sense if the tightness holds. On the DRAM side, TrendForce's October 21 commentary said DDR5 supply remains tight enough that suppliers are "clearly reluctant to sell" even as buyers raise their bids, while DDR4 stayed comparatively subdued as module makers ration capital. Micron's October 22 decision to abandon the 4F² DRAM cell architecture after twenty-plus years of research, sticking with the conventional 6F² design plus wafer bonding instead, doesn't add meaningful new bit supply on any near-term timeline. And China's CXMT reported October 19 that its own 4F² push is intact, projecting a nine-month net profit swing to as much as 132.5 billion yuan (roughly $19.8 billion) from a loss a year earlier, with its global DRAM share reportedly doubling to 10% over the past year per Counterpoint Research, even as its shares fell 7.8% on lingering doubts about the broader AI-capex cycle.

Driver: Unchanged since Issue 03: HBM production still eats disproportionately into capacity that would otherwise make conventional DRAM, and the same tightness has now shown up as hard numbers in a hard-drive maker's own quarter, not just in memory spot prices.

Watch for: A five-day earnings stretch starting October 26, when Rambus and Amkor both report, followed by Seagate on October 27, SK hynix's call on October 28, and Samsung and ASE Technology both on October 29. Seagate's number in particular is the next direct test of whether Western Digital's quarter (this issue's Second Cut) reflects an industry-wide HDD story or a company-specific one.

This storyline has taken three issues to play out, and it's worth retracing the arc before getting to the ending. Issue 09 covered Solidigm reportedly weighing a US IPO that could value it anywhere from $100 billion to $150 billion, a number that already looked aggressive for a company still without a US fab. Issue 11 found the deal moving forward anyway: Solidigm had named Goldman Sachs and Morgan Stanley as bankers for a reported listing, reopening a governance fight inside SK Group over how much of Solidigm's value should sit outside the parent company's own balance sheet. Last issue closed with Bloomberg Intelligence arguing, on October 18, that the funding benefits of a listing outweighed those governance concerns, while SK hynix's own position stayed carefully unresolved: no decision has been made. That changed within days.

SK hynix said October 22 that it is halting its review of pre-IPO capital raising and IPO options for Solidigm entirely, citing "recent capital market conditions" in its own statement. The real story sits underneath that phrase. South Korean financial regulators reportedly flagged that a US listing of Solidigm, a subsidiary SK hynix owns outright, could run afoul of the country's dual-listing rules, designed to stop a parent company from effectively spinning off value to outside shareholders without fairly compensating its own minority shareholders first. That's a materially different problem than market timing. Bloomberg Intelligence's funding-case argument from Issue 12 was a bet that SK hynix's board would weigh the dollars; the regulatory objection says the board may not have gotten to choose.

The stock move is the part worth sitting with longest. A 4.4% close, after an intraday drop of as much as 3%, means SK hynix shareholders spent the first part of that trading session bracing for the IPO to survive and the back half rewarding the company for killing it. That's the opposite of how markets usually treat a scrapped fundraising plan, and it's a much cleaner signal than anything in Issue 11's reporting about whether this deal was actually good for SK hynix shareholders versus good for unlocking Solidigm's standalone value. The answer, at least as judged by the people with money on the line, leaned toward "no."

What doesn't disappear is the problem the IPO was supposed to solve. Issue 08 covered Solidigm scouting a US NAND fab specifically to dodge export-control exposure tied to its South Korean and Chinese manufacturing footprint, and that fab still needs financing. SK hynix says it's now exploring "various funding options," which in practice means some combination of parent-company capital, debt, or a minority stake sale to a strategic partner, each of which keeps Solidigm's ownership structure, and SK Group's control over it, intact in a way a US listing never would have. That's probably the real takeaway: Korean chaebol structures resist diluting control even when a subsidiary's growth story would benefit from outside capital and the scrutiny that comes with it, and regulators this time gave SK hynix a clean excuse to make the choice its own governance instincts likely favored anyway.

Western Digital's fiscal Q1 2027 results, reported October 22 after market close, read less like a turnaround story and more like a company that simply stopped discounting. Revenue came in at $4.38 billion, up 55% year over year. Cloud revenue, 88% of the total, rose 54% to $3.9 billion on what management called durable demand for high-capacity nearline drives. The company shipped 248 exabytes, up 22% year over year and 8% quarter over quarter, while average price per terabyte climbed 28% year over year. Gross margin expanded to 58.1% and operating margin to 49.1%; non-GAAP EPS hit $4.64, up 161%. Guidance for the next quarter calls for 66 to 72% revenue growth. Volume production of the company's latest ePMR platform, good for up to 40TB per drive, is ramping, with HAMR technology still slated for the first half of calendar 2027.

Look at the relationship between those two growth rates and the real story shows up: exabytes shipped grew 22% year over year, but price per terabyte grew 28%. Western Digital isn't flooding the market with meaningfully more capacity than it was a year ago, it's charging noticeably more for roughly the same growth curve it already had. That lines up exactly with the 30TB-class enterprise HDD figure this newsletter has tracked since Issue 05, $1,216 in Q3 2026 against $495 in Q3 2025, a 146% jump that long predates this week's earnings call. The hard drive shortage isn't a volume story right now. It's a pricing story, and Western Digital's own guidance says it isn't done.

Seagate reports its own fiscal Q1 2027 results October 27, and 24/7 Wall St framed that print this week as a genuine test of two competing stories: Western Digital's blockbuster demand narrative against oversupply fears tied to Toshiba's capacity ramp, the same China fab upgrades and doubled AI-data-center HDD capacity this newsletter's between-issues log and Issue 11 both flagged. If Seagate's numbers echo Western Digital's, the "shortage runs for years" case Michael Dell made in Issue 12 gets another data point in its favor. If they don't, that's the first real crack in a narrative every HBM and NAND maker in this issue is currently pricing as settled.