The world's top supplier of high-bandwidth memory is reportedly weighing a bet that the numbers do not support. According to a Reuters exclusive on September 16, SK Hynix is in talks with Intel about a deal that would see it manufacture memory chips on US soil for the first time. Under one scenario it would lease part of Intel's long-planned Ohio campus. Under another it could form a venture with Intel and major cloud firms that are trying to lock in memory supply. Both companies pushed back on the specifics. SK Hynix said it is "reviewing various measures, including establishing additional production bases," but that "no matters have been determined at this stage." Intel called the report "speculation."
The real question is not whether a deal will close. It is why the most profitable memory business on earth would even consider building in the most expensive place to make chips, and what that says about who is actually pulling the strings.
The facts at list price
A few things are established, and they matter. SK Hynix is the leading supplier of HBM, the stacked memory that sits under the accelerators in AI systems. Its South Korea listed stock is up 400% over the last year, and in June it said it plans to double capacity over the next five years. In July it completed a secondary listing of American Depositary Receipts on the Nasdaq.
Its only US facility today is a packaging plant. On August 27, SK Hynix broke ground on an advanced packaging and R&D facility in West Lafayette, Indiana, at Purdue University. The company's own newsroom puts the investment at over $4 billion, with the cleanroom open by October 2028 and mass production of next generation HBM in the second half of 2029. The model is deliberate: cutting edge wafers are made in Korea, shipped to Indiana, then packaged and tested there before "Made in USA" products go to US customers. It is a packaging plant, not a wafer fab.
The Ohio talks would be a different animal entirely. Intel announced in 2022 that it would invest up to $100 billion to build what it called potentially the world's largest chipmaking complex in Ohio, with production originally scheduled for 2025. That has slipped. The two plants are now due to complete in 2030 and 2031, a delay reported in February 2025 and still in effect. The current two plant scope is roughly $28 billion.
| Facility | What it makes | Investment | Mass production |
|---|---|---|---|
| SK Hynix Indiana | HBM packaging and testing | over $4 billion | second half 2029 |
| Intel Ohio | leading edge logic wafers | about $28 billion (2 plant scope) | 2030 and 2031 |
The math the pitch leaves out
Manufacturing semiconductors in the United States costs materially more than in South Korea. Reuters' sources put the drivers at higher labour and construction costs, plus a supply chain that is still concentrated in Asia. That gap is not a rounding error. It is the difference between a business that prints record profits and one that has to be subsidised to break even.
The timeline makes it worse. The memory shortage is acute right now, in 2026. Our own analysis of the shortage, and why HBM is eating the wafer capacity that used to make ordinary DRAM, is here. A new Ohio memory line would not come online until 2030 or 2031, four or five years from now. A plant that opens in 2031 does not relieve a shortage that is squeezing server builders and PC makers in 2026. If the goal is to add supply for the current cycle, Ohio is the wrong tool.
So what is the Indiana plant actually for? It is the clever part of the whole story. By packaging wafers in the US rather than fabricating them, SK Hynix can produce "Made in USA" HBM by 2029 without the capital, the years of construction, or the cost premium of a full front end wafer fab. It satisfies the political requirement of domestic production while keeping the expensive, sensitive wafer work in Korea. That is the near term answer. The Ohio talks are the long term one.
The costs and caveats nobody mentions
The biggest obstacle is not money. It is Seoul. Advanced memory, HBM in particular, is treated as a sensitive national technology in South Korea. The trade ministry told Reuters that if a deal involves a "national core technology," it would be subject to review under the Industrial Technology Protection Act. At the same time, Seoul has been urging SK Hynix and Samsung to accelerate a new chipmaking cluster in the southwest of the country. SK Hynix is being pulled in two directions, and the two directions do not point the same way.
Then there is the leverage. US Commerce Secretary Howard Lutnick has threatened tariffs of up to 100% on South Korean and Taiwanese chipmakers that do not expand production in the United States. Reuters also reported that of a proposed $350 billion South Korean investment commitment made in return for lower US tariffs, about $150 billion has been earmarked for shipbuilding and the remaining $200 billion is undecided. Two of Reuters' sources said that has left SK Hynix in a difficult position, with the South Korean government seeking to use any US investment by the company as leverage in the talks while Washington pressures it to commit quickly.
Put plainly, the Ohio story is not an engineering decision. It is a bargaining chip in a trade negotiation, and the memory chips are the currency.
Our Read
The economics do not add up, and the companies that know this best are the ones hedging their language. SK Hynix is not going to build a full front end wafer fab in Ohio on cost. It will not do it for the current shortage, because the plant would not be ready until 2030 or 2031. What it will do, if it does anything, is mirror the Indiana model at a larger scale, or take on a slice of Intel's capacity that is cheap enough to justify and politically visible enough to satisfy Washington.
The near term supply answer is already on the ground in Indiana. That is the plant that can actually deliver US made HBM before the next cycle. The Ohio talks are about the cycle after that, and about who gets to claim the win. For Intel, a marquee memory tenant on the Ohio campus is the validation its foundry business has been waiting for, and that is why the shares jumped when the report broke. For SK Hynix, it is a way to keep the tariff threat at bay without betting the HBM franchise on a $28 billion bet in a state where the supply chain has to be rebuilt from scratch.
Watch two things. First, whether the South Korean trade ministry clears anything that touches HBM. That is the real gate, and it is the one most coverage misses. Second, whether the undecided $200 billion of the South Korean commitment gets tied to a memory commitment at all. If it does, the Ohio story stops being speculation and starts being policy. If it does not, this is a negotiating posture, not a roadmap, and the top HBM maker on earth will keep doing what it has always done, making the chips where it is cheapest and calling them American at the packaging stage.