SK Hynix's US Blueprint Hits a Paperwork Wall in Dalian

SK Hynix has spent the past year painting a picture of insatiable demand: memory sold out for years ahead, long-term contracts locking in capacity, and a US investment vehicle aimed squarely at the artificial-intelligence boom. What that narrative glosses over is a more mundane constraint — the machinery needed to build the plants still requires Washington's blessing, and that blessing has yet to arrive.

According to Reuters, the US Commerce Department has not granted requested approvals for American chipmaking equipment that SK Hynix needs to expand its facility in Dalian, China. The delay is not a rejection, but it carries consequences well beyond Chinese borders. SK Hynix reportedly expects to help fund part of its US expansion through improved productivity at Dalian — meaning the stalled permits could ripple directly into Solidigm's plans for a first US NAND plant at the Marcy Nanocenter in New York.

Two Projects, One Balance Sheet

The numbers attached to those plans are substantial. Media reports have floated an investment of up to 130 trillion won for the Marcy site, while Reuters described Solidigm's planned outlay of roughly $45 billion for two NAND fabs there. SK Hynix has stressed that the project remains under review and has not been finalized. Talks are ongoing.

Separately, SK Hynix is building HYAICO, a US-based investment arm focused on artificial intelligence, into which it has said it will pour up to $10 billion. A regulatory filing with the SEC indicates Chey Tae-won is slated to chair the entity. That figure represents a ceiling rather than money already committed — and the company has made no suggestion that HYAICO will bankroll the Solidigm fab. Investors would do well to keep the two ventures in separate mental columns.

Funding options for Solidigm narrowed on October 23, when the subsidiary shelved its review of pre-IPO and IPO alternatives, citing capital-market conditions and equity-price developments. SK Hynix said at the time that it continues to weigh further investment and financing avenues to support Solidigm's growth. The company has also said the subsidiary needs additional US production capacity to meet rising demand. The IPO pause, in other words, redirects attention to other funding routes rather than closing the growth chapter.

A Record Quarter on the Other Side of the Ledger

Against these open questions sits an unusually strong operating picture. SK Hynix reported third-quarter 2026 revenue of 104.6155 trillion won and operating profit of 83.3338 trillion won, crediting AI memory demand for the record figures. Long-term agreements cover more than 50 percent of production capacity for 2027–2028, according to the company, giving its planning a contractual floor.

The claim that DRAM and HBM output is effectively sold out through 2029 deserves closer reading. As media reports have noted, that assessment bundles secured long-term contracts together with customer demand still under negotiation. Not all of it is contractually locked in — a distinction that matters for anyone extrapolating from expected utilization to the reliability of future revenue.

The takeaway for shareholders is a split-screen valuation. Earnings power is not in dispute; execution risk on future capacity is. Whether customers want the memory is no longer the open question. What remains unresolved is when the equipment gets cleared, how the New York project gets financed, and which of those two answers arrives first.