Intel SK Hynix Memory Deal: What The Ohio Talks Could Mean

Intel (NASDAQ:INTC) shares jumped 4% after Reuters reported something that, for now, remains only a possibility. SK Hynix (NASDAQ:SKHY) is discussing ways to manufacture memory chips in the United States using Intel’s long-delayed Ohio campus. The Intel SK Hynix memory deal is therefore still at the discussion stage. One option involves SK Hynix leasing or using part of the site. Another involves a joint venture with Intel and major cloud companies. No transaction has been agreed, and even the type of memory that could be produced remains undecided. SK Hynix says it is reviewing different options, while Intel has declined to comment on what it calls speculation.

Still, there is a fascinating industrial logic behind these discussions. Intel has expensive U.S. manufacturing infrastructure and rising memory needs. SK Hynix has leading memory technology and customers asking for more supply. Cloud companies want secure AI-memory capacity, while Washington wants more semiconductor production on American soil. The complication is that Seoul also has a say in how sensitive Korean memory technology moves overseas.

Intel’s Ohio Problem & The Intel SK Hynix Memory Deal Could Become An Opportunity

Start with the strange reversal at the heart of this story. Intel once owned a major memory business. It then exited NAND by selling that operation to SK Hynix. Now, memory has become one of the constraints Intel must manage in the AI era.

That matters because Intel has another problem: a huge manufacturing footprint that needs better utilization. Its Ohio project was originally envisioned as a semiconductor complex that could eventually involve investment approaching $100 billion. The buildout has since been delayed as Intel has tied spending more closely to customer demand.

Intel Foundry’s financial results explain why utilization matters. In Q1 2026, the segment generated $5.4 billion of revenue but only $174 million came from external customers. It posted a $2.4 billion operating loss.

Conditions improved somewhat in Q2, but Foundry still lost about $2.1 billion.

This makes SK Hynix interesting. Intel has infrastructure, while SK Hynix has memory demand that it is struggling to fully satisfy. On its Q2 call, SK Hynix said customers were still requesting additional supply. Management expects tight memory conditions to persist for a considerable period. The Intel SK Hynix memory deal could therefore provide another route for addressing that capacity requirement.

Ohio could therefore connect two very different problems: Intel’s need to monetize manufacturing assets and SK Hynix’s need for more capacity.

The Structure & Memory Product Decide Everything

This is where investors need to be careful. A lease and a joint venture would tell two very different stories.

Under a leasing-style arrangement, Intel could mainly become a landlord. SK Hynix would gain access to U.S. infrastructure while Intel would improve utilization of an expensive campus. Strategically, however, Intel’s role in memory could remain limited.

A joint venture could go much further. Reuters reported that one scenario could include Intel, SK Hynix and large cloud companies seeking secure domestic memory supplies. Such a structure could spread the cost of U.S. production across companies with a direct interest in additional capacity.

The second unknown is just as important: what would actually be manufactured? Reuters reported that HBM, DRAM and NAND possibilities remain under discussion. No product has been selected. This means the Intel SK Hynix memory deal cannot yet be treated as an HBM project or tied to any specific memory product.

That distinction changes the economics and strategic significance of the project.

SK Hynix already began mass-production shipments of HBM4 during Q2. Management plans a broader production ramp during the second half. It also said HBM4 yields and quality were approaching mature HBM3E levels. Meanwhile, conventional server DRAM and enterprise SSD demand are benefiting from the same AI infrastructure buildout.

So investors should watch both variables together: ownership structure and product mix. Until those are settled, calling this an Intel-SK Hynix memory partnership gets ahead of the facts.

Seoul’s Technology Rules & Washington’s Manufacturing Push Could Collide

The geopolitical layer may be the hardest part to solve.

Washington has been pushing semiconductor companies to manufacture more chips inside the United States. SK Hynix is already moving in that direction. In August, it broke ground on an Indiana facility with planned investment exceeding $4 billion. The facility will focus on next-generation HBM and advanced packaging, with mass production targeted for the second half of 2029.

Ohio could represent another step. Yet moving sophisticated memory manufacturing abroad raises a different question for South Korea.

Reuters reported that South Korea’s trade ministry said HBM and DRAM technologies could fall under the country’s “national core technology” framework. That can trigger government review when sensitive technology moves overseas. This regulatory layer could therefore become an important consideration for the Intel SK Hynix memory deal.

That does not mean Seoul has decided to stop an Ohio project. The production decision itself remains SK Hynix’s responsibility, and no regulatory outcome has been determined.

SK Hynix itself was careful about overseas expansion during its July earnings call. Management said future locations would be judged using factors including power, water, labor, supply chains, semiconductor ecosystems and customer access. It also stressed that no additional overseas production decision had been made at that time.

The result is an unusual policy tension. Washington wants domestic semiconductor capacity, while Seoul has reasons to protect strategically important technology developed by Korean companies.

AI Demand Is Strong But The U.S. Cost Math Still Has To Work

The strongest argument for additional capacity is SK Hynix’s own demand picture.

Q2 revenue reached KRW79.3 trillion, up 51% sequentially. Operating income reached KRW60.5 trillion, producing a remarkable 76% operating margin. Management said AI infrastructure investment and tight memory supply continued to support pricing.

The demand story also extends beyond HBM. SK Hynix expects agentic AI to require more conventional server DRAM. Enterprise SSD demand is expanding as AI systems generate and process larger amounts of data. Its enterprise SSD revenue doubled sequentially in Q2, while Solidigm’s revenue from high-capacity enterprise SSDs more than tripled.

Customers are trying to secure that supply early. SK Hynix said it had completed long-term agreement negotiations with around ten customers. Typical agreements can run about five years and may include deposits and volume commitments.

That demand visibility makes a U.S. expansion easier to contemplate, but it does not automatically make it economical. Manufacturing in America can involve higher construction, labor and ecosystem costs than established Asian semiconductor hubs. Reuters identified those higher costs as a key issue surrounding the talks.

This helps explain why cloud participation could matter. Long-term customer commitments and outside capital could absorb part of the U.S. manufacturing premium. Without those protections, SK Hynix has to compare Ohio with expanding established production infrastructure closer to home. The economics of the Intel SK Hynix memory deal will therefore depend heavily on how those costs and capital commitments are ultimately shared.

Final Thoughts

There is plenty of strategic logic here, but there is still no signed Intel-SK Hynix deal. The eventual structure matters more than the initial headline.

A lease could help Intel monetize Ohio while giving SK Hynix additional U.S. capacity. A broader joint venture could carry much larger implications. It could reconnect Intel with the memory ecosystem it previously exited while bringing cloud customers directly into semiconductor capacity planning. But product choice, economics and South Korean technology review remain unresolved.

SK Hynix’s valuation also deserves context after its extraordinary earnings growth. As of September 16, the shares traded at roughly 6.59x LTM enterprise value/revenue, 8.69x EV/EBITDA and 7.71x LTM diluted earnings. The forward picture is substantially lower, including about 3.29x NTM EV/EBITDA and 4.33x normalized earnings.

Those multiples reflect a business whose earnings base has expanded dramatically with the AI-memory cycle. They also leave investors weighing how durable today’s unusually strong profitability will prove as SK Hynix commits more capital to future capacity.

For now, the Intel SK Hynix memory deal and the Ohio possibility do not change that valuation story by themselves. The important question is whether these exploratory talks eventually become a structure that improves SK Hynix’s supply position without undermining investment returns. Until the parties decide who pays, what gets produced and what Seoul permits, the market is reacting to strategic possibility rather than a completed transaction.

Disclaimer: We do not hold any positions in the above stock(s). Read our full disclaimer here.