Solidigm Plans to Build First U.S. NAND Wafer Fab in New York State
nashnova research
SK Hynix subsidiary Solidigm is weighing a first-ever US NAND flash wafer fab in upstate New York, which would give it a second production base beyond its Dalian, China plant — but no deal has been reached and key details remain under discussion.
Why upstate New York?
Reuters, citing people familiar with the matter, reports that upstate New York is the leading candidate site — but no agreement has been signed.
Solidigm's only NAND fab today sits in Dalian, China. This means → its entire flash production is exposed to a single geography and a single set of geopolitical risks.
In plain terms = the core logic is not capacity growth but risk diversification — securing a production node that does not depend on Dalian.
How many US manufacturing lines does SK Hynix already have?
Indiana is furthest along: ground was broken in August, total investment exceeds $4 billion, cleanroom activation is planned for October 2028, and next-gen HBM volume production is targeted for H2 2029.
But Indiana handles packaging and testing — front-end HBM wafers will still be made in Korea and shipped over. This means → it addresses back-end capacity, not front-end manufacturing.
Ohio is far more tentative: SK Hynix is in early talks with Intel about leasing facilities or forming a joint venture with Intel and major cloud companies, but what chip would even be produced there is undecided. Intel's two planned Ohio fabs have been pushed to 2030 and 2031.
Solidigm's New York project and the Ohio talks are independent of each other — three separate tracks.
What underpins the case for NAND expansion?
Solidigm holds a clear competitive edge in high-capacity QLC enterprise SSDs — flash drives that store data with four charge levels per cell, yielding large capacity at lower unit cost.
This reflects a broader trend: data centers are scaling compute and storage simultaneously, driving sustained enterprise SSD demand.
SK Hynix is shifting its NAND portfolio toward higher capacity and performance. In Q2 this year, 321-layer products became the largest node by NAND output share; the target is to raise that to roughly 50% of Korean NAND capacity by year-end.
How does customer lock-in shape the expansion pace?
At group level, SK Hynix has signed long-term agreements with roughly 10 customers, including core strategic partners.
This means → expansion is not speculative — it is paced to committed demand, with investment staged against order visibility.
Negotiations with other major customers are ongoing; future investment will be phased according to customer needs and capital efficiency.
What is the biggest obstacle to building in the US?
Cost is the central challenge: US labor, construction, and supply-chain costs are significantly higher than Korea's, and the semiconductor supply chain remains heavily concentrated in Asia.
The competitive landscape is also shifting: China's DRAM maker CXMT (长鑫存储) is setting up a pilot NAND line at a new Beijing facility, potentially emerging as a new supplier alongside Samsung, SK Hynix, and Micron.
In plain terms = SK Hynix faces a straightforward trade-off: is the strategic value of US-based production worth the premium cost? The answer is still open.
市场有风险，内容仅供研究参考，不构成投资建议。