
Solidigm IPO: Why SK Hynix's $150bn NAND Listing Must Come Early in 2027
SK Hynix wants a U.S. multiple for a flash-memory business it bought from Intel for $8.8 billion. On current earnings the price is defensible. On timing it is not: a listing in late 2027 would be priced off 2028 earnings, deep into a memory upcycle that is already more than a year old.
SK Hynix is looking at a straightforward valuation arbitrage. Its own shares trade at about eight times trailing earnings, while Sandisk, the only large U.S.-listed pure play in NAND flash, trades at 24 times. Solidigm, the Californian enterprise-SSD unit SK Hynix built out of Intel's old flash business, is considering a U.S. listing as early as next year that could value it at up to $150 billion and raise about $15 billion, Reuters reported on Friday. Banks pitched for roles this week. That price can be justified today. SK Hynix should list in the first half of 2027, or accept a number closer to $100 billion.
What the buyer gets
Solidigm's earnings are recent but substantial. Its holding company reported first-half revenue of 12.25 trillion won and net profit of 5.84 trillion won, about 44 times the year-earlier figure, according to results reported by Korean business media in August. At about 1,390 won to the dollar, that is roughly $8.8 billion of revenue and $4.2 billion of profit in six months. Annualized, $150 billion comes to about 18 times earnings and 8.5 times sales.
Sandisk makes that valuation look restrained. Sandisk is worth $260 billion on trailing revenue of $20 billion and an operating margin of 62%, which works out to almost 13 times sales, according to StockAnalysis data. Solidigm also has a better position in the part of the market with the most demand. TrendForce says enterprise-SSD revenue at the top five suppliers doubled in the second quarter to $37.6 billion. SK Hynix's group ranked second, with Solidigm's high-capacity QLC drives doing much of the work, according to TrendForce. Those are the dense drives cloud operators put next to GPU clusters to hold training data and inference caches.
Onshore supply is worth more in the current trade climate. Earlier this month Reuters reported that Solidigm is considering a NAND fab in the United States, with upstate New York the leading candidate. An IPO that raises $15 billion would go a long way toward paying for such a plant, and would give a Korean-owned supplier a U.S. listing, a U.S. factory and a U.S. board. That has strategic value beyond the multiple.
Why the timing matters more than the multiple
Forward multiples show where the market's doubts lie. Sandisk trades at about 8 times forward earnings, Kioxia at about 4 times and SK Hynix at about 4.6 times, all well below their trailing multiples. Analysts expect profits to rise sharply, and investors are refusing to pay for them. That is how markets treat memory makers near the top of a cycle: the higher the earnings, the lower the multiple.
The chart shows why. Kioxia's shares are up more than tenfold in a year and Sandisk's about seventeenfold. Kioxia listed in Tokyo in December 2024 after postponing its offering because of the weak market. Its value has since risen to about 30 trillion yen. Timing, more than anything management did, explains that return.
Solidigm's price also depends on earnings doubling. At Sandisk's forward multiple, $150 billion needs about $18 billion of annual profit, more than twice the first-half run rate. Enterprise-SSD contract prices are still rising, and TrendForce expects that to continue in the fourth quarter on North American demand. The investors buying a 2027 IPO will be pricing 2028. Rising output from Chinese suppliers, which TrendForce names as a risk, has more time to arrive by then. So does the capacity the current profits are paying for, including, in time, Solidigm's own U.S. fab.
The bull case and its limit
The bull case says AI has made NAND a structural business instead of a commodity. Sandisk has signed long-term supply agreements worth $93.9 billion covering about two-thirds of its fiscal 2028 output. If Solidigm can show similar contracts, investors will treat its earnings as secured backlog, not cyclical spot sales.
Take-or-pay contracts reduce volume risk, but they are repriced over time. They also have not been tested in a downturn, when customers who signed in a shortage have every reason to renegotiate. Buyers will pay for visibility, but a single good contract year will not earn Solidigm the multiple of a software company. The contracts could narrow the discount to peers. They are unlikely to remove it.
The Seoul problem
There is a second group of shareholders to satisfy. Korean media have already called the plan a "quadruple listing": SK Inc owns part of SK Square, which owns part of SK Hynix, which would float Solidigm. Duplicate listings are a live political issue in Korea, where minority shareholders at a parent watch value move down to a new layer of the structure. SK Hynix's statement that Solidigm "is reviewing various options" and that "no specific plans have been confirmed" leaves room for both.
SK Hynix has a reasonable defence. Its shares, at roughly 1.86 million won, are valued at about 1,357 trillion won, according to StockAnalysis. A $150 billion subsidiary would re-rate the parent's NAND business, which is priced today as an afterthought next to its high-bandwidth DRAM. If the IPO puts a U.S. price on assets Seoul is ignoring, parent shareholders gain. The risk is a price that later falls and drags the parent's valuation down with it.
The call
SK Hynix paid $6.6 billion in 2021 and $2.2 billion this year to complete the Intel purchase. A listing at $150 billion would be one of the best returns on a semiconductor acquisition in memory. It is available only while investors still extrapolate the current cycle.
We expect Solidigm to file confidentially with the SEC by the end of the first quarter of 2027 and to aim for a listing before mid-year. A deal completed in that window should clear between $120 billion and $150 billion. If the offering slips to the second half, when investors will be setting 2028 earnings against more NAND capacity coming online, the price is more likely to land nearer $100 billion. For a business that sells a cyclical product, the listing date is the valuation decision.