The underwriters are chosen, but nothing is confirmed
Bloomberg reported on September 7th that SK Hynix has selected Goldman Sachs and Morgan Stanley as lead underwriters for a US initial public offering of Solidigm, its American NAND flash subsidiary, with JP Morgan, Citigroup and UBS also joining the syndicate. Preparatory work for a pre-IPO fundraising round is said to be under way. The appointment of a full banking syndicate alongside pre-IPO discussions signals that preparations have moved meaningfully forward.
That said, no decision has been taken. On October 1st, SK Hynix reiterated that it was "reviewing various options related to Solidigm, with nothing confirmed," emphasising that any decision would be guided by long-term shareholder value. The numbers, too, remain unsettled. Reuters, citing multiple sources, reported on September 25th that the offering could raise around $15bn at a valuation of up to $150bn (roughly 204 trillion won); Bloomberg, both last month and again this week, put the figures at $10bn raised and a valuation of up to $100bn. All figures come from anonymous sources and could change. The earliest plausible listing date is 2027.
Why domestic investors have little to cheer
The market's initial reaction was negative. On September 28th, the first trading day after Bloomberg's report, SK Hynix fell 5.05% to 1,768,000 won in regular trading, while SK Square dropped 7.56% to 1,100,000 won. Context matters, however: Samsung Electronics fell 5.43% on the same day, and the KOSPI — South Korea's benchmark index — declined 2.7%, moves attributed in part to rising oil prices stoking risk aversion. The Solidigm news was therefore not the only force at work.
The structural concerns, however, are real. SK Hynix sits at the bottom of a four-tier listed-company stack — SK Inc. owns SK Square, which owns SK Hynix, which in turn controls Solidigm. Floating a fourth entity in this chain risks diluting the value of the NAND business as reflected in SK Hynix's own share price. Issuing new shares would also hand a portion of future growth to outside investors. SK Hynix already has American depositary receipts trading on Nasdaq, making the ownership structure still more complex.
The strategic rationale has also been poorly communicated. An analysis of SK Hynix's interim financial report by the Herald Economy found that Solidigm's NAND division recorded revenues of 12.25 trillion won in the first half of this year — more than double the 5.82 trillion won of the preceding six months — driven by surging AI data-centre demand and sharply higher NAND prices. SK Hynix's parent operations generated revenues exceeding 131 trillion won in the same period. Against this backdrop, the need for external capital is far from obvious. The company's only public explanation has been that "investing entirely from internal funds is not necessarily optimal."
Regulatory conditions have also tightened. Guidelines on dual listings introduced in August impose five obligations on parent-company boards — including shareholder impact assessments and protective measures — and apply equally when a subsidiary lists on a foreign exchange. Non-compliance can result in a penalty of up to 1bn won and a one-day trading suspension. The Financial Services Commission has noted that overseas listings bypass the Korea Exchange's prior review, making penalties more likely. How these rules apply to a great-grandchild entity such as Solidigm remains legally ambiguous. The minority-shareholder advocacy group Act has asked the exchange for a formal ruling and has complained that shareholders cannot tell what the board is actually deliberating. Some analysts argue that Solidigm is different in kind from the "spin-off listing" phenomenon that the guidelines target, given that it is an externally acquired asset funded in dollars. Others see a genuine upside in unlocking hidden value. But without disclosure of the new-to-old share ratio and post-listing ownership stakes, the net benefit cannot be calculated. The biggest gap, ultimately, is one of communication.
If not money, then Dalian
The more revealing clue may lie in geography. Solidigm is incorporated in California, but its wafer fabrication is concentrated in Dalian, in north-eastern China. The company has no manufacturing presence in the United States — only research, development and sales operations. According to the Seoul Economic Daily, the Dalian plant produces primarily general-purpose NAND at around the 100-layer node, largely based on designs inherited from Intel, while higher-end products of 300 or more layers are made in South Korea, centred on the Cheongju complex.
Regulatory risk has already flared once. In August last year, the US Department of Commerce announced it would revoke Verified End User status — which allows approved companies to import controlled equipment into China — for Samsung's Xi'an plant and SK Hynix's Wuxi and Dalian facilities. A last-minute negotiation converted the arrangement to an annual approval system. The immediate crisis was averted, but equipment imports for capacity expansion and process upgrades remain blocked, and approvals must be sought each year.
What makes the picture more complicated is that Solidigm appears to be expanding in China, not retreating. The Seoul Economic Daily reported in August that Solidigm had this year resumed investment in Dalian's second fabrication plant, which had been idle for some time. Adding a new line of roughly 50,000 wafers per month to the existing first plant, which produces approximately 100,000 wafers per month, would expand local NAND capacity by around 50%. Equipment installation could begin as early as November, with mass production targeted for the first half of next year. How this investment is being squared with American export controls is not clear from available reporting. The pull is obvious — booming demand for NAND in AI servers creates strong incentives to expand even commodity capacity — but it also means that China dependency could increase before it diminishes. Expansion in Dalian and exploration of a new American foothold appear to be proceeding simultaneously.
What Washington wants: build locally or pay tariffs
The American pressure is explicit. Commerce Secretary Howard Lutnick told CNBC on September 2nd that companies building semiconductor manufacturing facilities in the United States would receive tariff relief, while those that did not would pay tariffs. The United States imposed 25% tariffs on certain advanced AI chips in January, designating that move as a "first step," with broader application signalled to follow. Neither Samsung nor SK Hynix has front-end wafer fabrication for DRAM or NAND on American soil. SK Hynix is building an HBM packaging facility in Indiana at a cost of more than $4bn, but the wafers themselves are made in South Korea. Whether tariff relief would be assessed at the company level or product-by-product against specific domestic production facilities has not been determined. A product-specific standard would favour Micron, which is breaking ground on a new wafer fab in New York, and would intensify pressure on Korean chipmakers to establish front-end fabs in America.
This is where Solidigm's position becomes strategically interesting. ZDNet Korea reported that Solidigm is considering establishing its first NAND production facility on the US East Coast, with New York state mentioned in some accounts. Site, scale, investment size and timing are all undecided, and a Solidigm spokesman confirmed that "nothing has been finalised." The move, if it proceeds, is expected to be underpinned by long-term supply agreements with major American technology customers. In financial circles, the logic is straightforward: raising dollars from American investors to build an American factory is a pre-emptive hedge against tariffs and geopolitical risk — and a politically persuasive one in Washington.
The Japan card: Kioxia and a possible new base
Japan is also part of the picture. In August, Kioxia disclosed that BCPE Pangea Cayman 2 — a special-purpose vehicle created by Bain Capital that holds a 14.19% stake — had overtaken Toshiba (14.12%) to become its largest shareholder. SK Hynix holds convertible bonds issued by that vehicle, having invested 129bn yen when it joined Bain's consortium in the 2018 acquisition of Toshiba's memory business. Some reports have described SK Hynix as the "de facto largest shareholder" of Kioxia as a result.
That characterisation warrants caution. The convertible bonds have not yet been exchanged for ordinary shares, and SK Hynix has stated that no internal discussions about conversion have taken place. Converting and exercising voting rights would require competition-authority clearance in multiple jurisdictions, with Japan's review considered the most demanding hurdle. SK Hynix also committed, at the time of its 2018 investment, to keeping its voting stake in Kioxia below 15% until 2028. Nevertheless, the arithmetic is notable: according to Counterpoint Research, first-quarter NAND market shares this year stood at Samsung 29%, SK Hynix 18% and Kioxia 14%. Combined, SK Hynix and Kioxia would theoretically lead the market.
Broader Japan ambitions are circulating. Chey Tae-won, chairman of the SK Group conglomerate, has publicly cited Japan as a candidate for the group's next production hub. Industry analysts are speculating about possibilities ranging from a joint NAND production arrangement with Kioxia to a new site for DRAM or advanced packaging. None of this is confirmed. Kioxia's separate plan, announced alongside Western Digital (owner of the SanDisk brand), to invest more than 5 trillion yen by 2032 is a distinct commitment that does not involve SK Hynix.
Four locations, one unanswered question
Assembling these pieces produces a coherent, if speculative, map. South Korea handles leading-edge NAND of 300-plus layers. China's Dalian supplies general-purpose product. A new American plant would reduce tariff and political exposure. Japan offers a potential partnership with Kioxia and a further production base. In this reading, the Solidigm IPO is primarily a vehicle for financing the American leg of that strategy.
Yet the framing is not quite "decoupling from China." Solidigm is simultaneously expanding in Dalian. The more accurate description is a structure that preserves Chinese manufacturing while adding an American buffer — a hedge, not a pivot.
The limitations of this thesis are equally clear. First, a stock-market listing does not insulate a company from export controls or tariffs. Those depend on where factories are located and who controls them, not on where shares are traded. If SK Hynix retains a dominant stake in Solidigm after listing, the company's American identity will be a thin shield. Second, SK Hynix has never stated geopolitical diversification as a rationale for the IPO. Third — and most importantly — a strategy that is not explained to shareholders is, from their perspective, no strategy at all. Until the company presents a coherent account that encompasses the Dalian risks, the scale of the American plant and the Japan ambitions, the market will keep asking the same question: why now, and why external capital?
What to watch
Several developments will clarify the picture. The first is disclosure of the offering structure — specifically the ratio of new to existing shares and SK Hynix's post-listing ownership stake in Solidigm. The second is whether the proposed American NAND factory acquires a confirmed site and scale. The third is whether Washington sets tariff-relief criteria at the company level or ties them to specific product-line domestic production, a distinction that could determine whether the American listing provides meaningful commercial protection.
Beyond these, observers should track whether equipment deliveries for Dalian's second plant proceed on schedule — tentatively from November — and how that process interacts with the annual US approval cycle. The progress of any regulatory review of SK Hynix's Kioxia convertible bonds, and the seriousness of the Japan production hub discussions, will fill in further pieces. Finally, how SK Hynix's board handles its obligations under South Korea's new dual-listing guidelines — and whether it chooses to communicate transparently with shareholders — may prove to be the decisive factor in whether the Solidigm flotation succeeds at all.