SK hynix subsidiary Solidigm is in talks with the US Department of Commerce about building a $45 billion NAND flash memory plant in the United States. Bloomberg reported on October 27, 2026 that Solidigm is considering a US investment while also seeking export licenses for the semiconductor manufacturing equipment it needs to upgrade the technology at its existing fab in Dalian, China. Just four days earlier, parent company SK hynix had announced that it was ending its review of a pre-IPO funding round and an initial public offering (IPO) for Solidigm. The sizable investment proposal has surfaced as funding options are being reassessed, but making it happen will require not only securing construction funds but also dealing with US export controls. At this point, neither the investment amount nor the construction plan has been formally decided, so the company's investment decision and the government's approval need to be considered separately.
A $45 Billion NAND Plant in the US, With Investment Possibly Growing to $100 Billion
Bloomberg's report is based on anonymous sources familiar with the matter. The $45 billion figure is the scale of investment Solidigm is said to have presented in its talks with the Commerce Department; construction has not been formally approved. Likewise, the export license for equipment needed to upgrade the China fab is not something Solidigm is said to have obtained in exchange for a US investment. It is one of the conditions being sought in negotiations. As of now, neither the investment plan nor the export license has been finalized.
According to the sources, Solidigm is also considering a plan to upgrade the Dalian fab in China and use the revenue it generates to fund construction of the new US plant. Over the long term, US investment could reach as much as $100 billion, and New York State is among the candidate sites. Construction could begin as early as 2027, but all of this is information at the consideration stage. The figure of up to $100 billion also only indicates the potential for future expansion; additional investment on top of the $45 billion has not been committed.
Even if construction does start in 2027, the completion date and the start of mass production have not been set. NAND flash memory, the subject of this project, is non-volatile memory that retains data even when power is off, and it is the semiconductor that provides the storage capacity in SSDs. Building a plant to manufacture NAND chips differs from expanding facilities that assemble finished SSDs, in both process and investment content. The report does not reveal specifics such as the number of NAND layers to be used or the wafer production capacity, so actual supply capacity cannot be estimated from the $45 billion investment figure.
IPO Review Ends, $45 Billion Proposal Emerges: Funding and Export Licenses Are Separate Issues
In a statement released on October 23, SK hynix said it had ended its review of a pre-IPO round and an IPO for Solidigm, citing factors such as market conditions and share prices. It maintained its previous view on the need for additional production capacity in the US, however, and said it would continue to evaluate the timing, scale, and funding method of any investment. In other words, what has ended is the consideration of raising funds through the equity market; capital investment in the US itself has not been abandoned. At the same time, there has been no announcement that another funding method has been formally adopted, so there is no basis for concluding that the $45 billion in construction funds has been secured.
Furthermore, whether construction funds for the US plant can be secured and whether export licenses for the equipment needed at the China fab can be obtained are different questions, with different decision-makers and different conditions. The former mainly concerns the investment and financial decisions of Solidigm and SK hynix, while the latter depends on the US government's export control policy. To understand this investment concept, it is necessary to sort out what each announcement and report establishes and what remains undecided.
| Date / Source | Subject | What can be confirmed | What remains undecided |
|---|---|---|---|
| Oct. 27, 2026 – Bloomberg | Investment in a US NAND plant | Report that $45 billion investment is being discussed with the Commerce Department | Construction approval, final investment amount, securing of funds |
| Oct. 23, 2026 – SK hynix | Pre-IPO/IPO and additional capital investment | Review of pre-IPO and IPO ended. Maintains the need for additional capacity in the US; continues to evaluate timing, scale, and funding method | Adoption of alternative funding method, decision to build the US plant |
| Oct. 1, 2026 – SK hynix | Internal funds and investment allocation | Internal funds are also an option; investment in HBM, server DRAM, and enterprise SSDs to be judged comprehensively | Final allocation of funds to Solidigm, whether everything will be funded internally |
| Aug. 29, 2025 – US Commerce Dept. BIS | Equipment exports to China fabs | Intends to grant export licenses to maintain existing operations of former VEU participants. However, no approval for capacity expansion or technology upgrades | Decision on this specific application, whether any exception accompanies the US investment |
Sorting things out this way shows that whether the $45 billion proposal comes to fruition does not depend solely on whether Solidigm can raise a huge sum. In addition to the parent company's investment decisions and funding allocation, what kind of equipment exports the US government allows for the China fab is also an important condition. The idea Bloomberg reported, of using Dalian fab revenue for US investment, links these two issues. However, SK hynix's official statement does not confirm the $45 billion investment figure, and the specific license conditions the US government may have presented in these talks have not been made public.
The Idea of Using Revenue From the Dalian Fab to Fund a New US Plant
Solidigm was created when SK hynix acquired Intel's NAND and SSD business, and its main business today is high-capacity enterprise SSDs for AI data centers. A key point in considering this US investment proposal is which businesses the parent company directs its funds toward. In its October 1 explanation, SK hynix said that investment with internal funds is among the options, but that even with ample cash, funding everything internally is not necessarily the best approach. The company plans to judge investment in HBM, server DRAM, and enterprise SSDs comprehensively, so the question is not only whether funds can be secured but also where to allocate limited management resources.
In its October 23 statement as well, SK hynix said that existing production capacity alone would struggle to keep up with growing AI-related memory demand and that additional production facilities would be needed in the US. At the same time, it said it would keep evaluating the timing, scale, and funding method of investment, meaning business necessity does not translate directly into a decision to build. With investment also needed in HBM and server DRAM, how to secure a large sum for Solidigm becomes an issue for the entire group's capital investment strategy.
Overlapping with this is the idea Bloomberg reported of linking the Dalian technology upgrade with US investment. An IPO is a way to raise capital from outside investors, whereas revenue from Dalian operations is money generated through business activity, so the two are different in nature. Export licenses for China-bound equipment do not themselves generate funds; they are a condition for maintaining and upgrading the fab's production equipment. Therefore, even if the Dalian technology upgrade were approved, that alone would not secure construction funds for the US plant.
In the first place, figures such as how much additional revenue the Dalian upgrade could bring, and what share of it could go to US investment, have not been disclosed. It also cannot be confirmed that US investment would be impossible without the technology upgrade, or that ending the IPO review has left Solidigm short of funds. Since SK hynix is examining multiple options, including internal funds, using the Dalian fab to secure funding should for now be regarded as one reported idea.
US Equipment Export Controls at the Center: Existing Operations and Technology Upgrades Are Treated Differently
A major potential obstacle in these talks is US export controls on semiconductors to China. In its August 29, 2025 announcement, the Commerce Department's Bureau of Industry and Security (BIS) revoked the Validated End-User (VEU) exemption, which had allowed certain manufacturing equipment to be exported without individual licenses to semiconductor fabs owned by foreign companies in China. BIS indicated it intends to permit exports of equipment needed to maintain existing operations at China fabs of former VEU participants. However, it stated explicitly that it would not permit equipment exports intended to expand production capacity or upgrade manufacturing technology.
This distinction carries real weight for Solidigm's Dalian fab. Under the US government's published policy, exporting equipment needed to keep existing facilities running is treated differently from exporting equipment to introduce newer manufacturing technology. If the Dalian technology upgrade Bloomberg reported falls into the latter category, a policy of permitting exports to maintain existing operations alone would not allow Solidigm to obtain the equipment it needs. That said, the policy BIS laid out in 2025 is not a formal decision on this individual negotiation. Nor has it been decided that the government will make an exception for technology-upgrade equipment exports on the condition of a large US investment.
If an export license for Solidigm's China fab is reported in the future, its intended use will need to be checked carefully. Whether it is limited to maintenance and keeping existing equipment operating, or extends to manufacturing technology upgrades and capacity expansion, will greatly change the impact on this investment concept. For the US plant, too, it matters how concrete Solidigm and SK hynix make the construction schedule, final investment scale, and funding method.
As demand for SSDs for AI data centers grows, a new NAND plant, if built, could affect future supply capacity. However, what has come to light only goes as far as a proposal to break ground in 2027 at the earliest, and neither the start of mass production nor wafer production capacity has been indicated. It is not yet a stage at which the $45 billion investment figure alone can be used to predict the impact on NAND supply or SSD prices.
What deserves attention in this report is not only the size of the investment but also the fact that expanding semiconductor production in the US and upgrading the China fab are tied together in a single negotiation. As Solidigm's investment decision and the US government's export license conditions become more concrete, it will be possible to judge more accurately whether the $45 billion plan can move toward actual plant construction.