NextFin News - SK Hynix is in talks with Intel about a deal that would see it manufacture memory chips in the United States for the first time, sending shares of both companies higher in premarket trading on Wednesday, while J.B. Hunt Transport Services fell as much as 9% after the trucking company warned that third-quarter earnings will decline 5% to 10% from the second quarter on rising operating costs. The two moves frame the same AI-era tension from opposite ends: one company is being asked to onshore capacity at a premium, the other is watching its margin get squeezed by the inputs that onshoring makes more expensive.
The Deal Under Discussion
Under one potential scenario, SK Hynix would lease part of Intel's long-planned chipmaking facility in Ohio; under another, it could form a venture with Intel and major cloud firms seeking to lock in memory supplies, according to three people familiar with the discussions. The talks are exploratory, one source said, and no decisions have been made. Intel declined to comment on what it called speculation, adding only that it is continuing with investments in Ohio to ready the site. SK Hynix said it is reviewing various measures, including establishing additional production bases, to strengthen the competitiveness of its memory business, but that no matters have been determined at this stage.
The stakes are substantial. Intel's Ohio complex, first announced in 2022, is planned to cover roughly 1,000 acres and hold up to eight fabrication factories, with a long-term investment that could exceed $100 billion. But the site's initial production has been delayed into the 2030-2031 time frame, and a SK Hynix tenant — or a joint venture that brings cloud customers onto the site — would be one of the few ways to put that capital to work sooner.
"We're close to closing some deals that are in the billions of dollars per year in terms of revenue," Intel CFO Dave Zinsner said at the Morgan Stanley TMT conference in March, referring to the foundry division's advanced-packaging pipeline.
The deal would relieve pressure on Intel, which has been struggling, and mark a big win for the Trump administration, which has been ramping up pressure on chipmakers to build in the United States as relentless investment in AI and data centers fuels an acute shortage of memory chips. But potential opposition from Seoul could be a major hurdle. South Korea's trade ministry said any decision would be at the company's discretion, but if a project involved national core technology it would be subject to review under the Industrial Technology Protection Act. Manufacturing semiconductors in the United States costs much more than in South Korea, two of the sources said, citing higher labor and construction costs and an Asian-centered supply chain.
Why This Is Different From the Packaging Rumors
This is not the same story as the recent reports that SK Hynix is testing Intel's 2.5D EMIB packaging technology for high-bandwidth memory. Those reports, which sent both stocks sharply higher earlier this week, concerned a supply-chain diversification move at the packaging stage. The talks reported today go further: they concern front-end memory fabrication on U.S. soil, which is both strategically more significant and politically more sensitive.
The distinction matters because it changes what is being negotiated. Packaging is a capacity bottleneck that customers feel today; front-end memory production is a geopolitical commitment that governments feel. SK Hynix recently completed a secondary listing on the Nasdaq in July, raising roughly $26.5 billion at $149 per American depositary share, and it controls an estimated 58% of the high-bandwidth memory market. Washington's interest in anchoring that capacity in the United States is a direct function of that dominance. The same dominance is why Seoul would scrutinize any deal involving HBM or DRAM as strategically sensitive technology.
This is a structural shift in the semiconductor supply chain, not a cyclical rumor trade, for three reasons. First, the demand driver is physical and persistent: AI accelerators require ever more high-bandwidth memory stacked on logic, and the shortage is bleeding into consumer electronics. Memory contract prices surged 80% to 90% in the first quarter of 2026 alone, according to Counterpoint Research, after rising 50% in the prior period. Device makers are passing those costs through: Microsoft's latest Surface Pro models start at $1,599, $600 more than the previous generation, and Nintendo lifted the Switch 2 price by $50 to $499 in May. Second, the capacity math is structural: a research note from Morgan Stanley forecasts global demand for TSMC's CoWoS packaging wafers to reach 1 million units by 2026, with Nvidia alone absorbing roughly 60%. Customers that cannot get CoWoS capacity have a standing incentive to qualify a second source, and governments that depend on AI infrastructure have a standing incentive to onshore it. Third, SK Hynix's own capital commitments show it is already treating U.S. presence as strategic: it broke ground last month on a $4 billion HBM facility in Indiana, and approved a ₩19 trillion ($12.9 billion) packaging and test facility in Cheongju, South Korea, in January.
Even so, the market is right to treat the reports with caution. SK Hynix denied a separate rumor in July that it was pursuing an acquisition of Intel's Ohio site, and the company has now said nothing has been determined on the latest talks. Exploratory discussions are not a supply agreement, and Intel's foundry business has a history of promising customer wins that arrived later than expected. The structural case does not depend on this specific deal closing, but the stock reaction does.
There is also a harder constraint that any Intel-SK Hynix arrangement must eventually confront: process technology. More than 90% of Intel's next-generation Nova Lake desktop CPUs are expected to be manufactured on TSMC's N2 process rather than Intel's own fabs, according to industry reporting, and the company's 18A node faces its first commercial stress test with Panther Lake laptops shipping later this year. If Intel cannot yet run its own flagship desktop silicon in its leading-edge fabs, convincing SK Hynix to place advanced memory production there is a heavier lift than hosting a lease. The Ohio site's delayed timeline — initial production pushed into 2030-2031 — cuts both ways: it gives Intel time to prove the node, and it gives SK Hynix time to decide it would rather wait for TSMC's Arizona fabs or expand in Indiana.
JB Hunt: The Other Side of the Same Inflation
While chip investors cheered a potential onshoring breakthrough, freight investors sold first and asked questions later. J.B. Hunt said at the Morgan Stanley Laguna Conference that it expects earnings to decrease 5% to 10% sequentially from the second to the third quarter, a departure from its usual practice of not providing in-quarter updates. The company did not issue a revised earnings-per-share forecast or a detailed breakdown of the anticipated cost increases.
The tension is visible in the company's most recent results. In the second quarter, J.B. Hunt reported revenue of $3.50 billion, up 19% from a year earlier, and operating income of $259.5 million, up 32%. Earnings per share came in at $1.91 versus $1.31 a year earlier. Intermodal was the standout, with revenue up 22% to $1.75 billion and operating income up 58% to $150.9 million. The Integrated Capacity Solutions segment swung to a $1.7 million operating profit from a $3.6 million loss a year earlier on revenue up 49% to $388 million. But the Truckload segment, despite a 35% increase in revenue to $240 million, posted a $1.3 million operating loss as higher purchased transportation costs outweighed gains.
That is the crux of the premarket selloff. A Barclays analyst noted that the company indicated freight demand appeared to have begun improving, but that higher costs were expected to affect near-term profitability, identifying increased fuel prices and associated surcharges as likely contributors. In other words, the top line is turning, but the cost line is turning faster.
The fuel backdrop explains why. The average less-than-truckload fuel surcharge in the second quarter of 2026 surged to more than 60% above the June 2025 level, driven by diesel prices that were 51% higher year over year. DAT Freight & Analytics reported that the spot van rate reached $2.52 per mile in March, up 53 cents from a year earlier, while the spot reefer rate rose 70 cents to $2.97 and the flatbed rate climbed 56 cents to $3.09. Rates are rising — which should help a carrier's revenue per load — but the fuel bill is rising faster, and the surcharge mechanism that is supposed to bridge the gap lags on the way up.
Cyclical Squeeze, Not a Broken Model
JB Hunt's pressure is cyclical, and it should mean-revert, because the driver is input-cost inflation layered on a freight cycle that is still in early recovery rather than a permanent loss of pricing power. Fuel surcharges are a pass-through mechanism by design; they lag spot fuel prices on the way up and on the way down, which creates exactly the kind of sequential margin compression the company flagged. Once fuel stabilizes, the surcharge mechanism stops working against the carrier.
The evidence for a cyclical rather than structural read is in the segment mix. Intermodal demand is being driven by customers facing higher fuel prices and constrained capacity in other modes — the same fuel spike hurting Truckload is pushing freight toward rail. Dedicated Contract Services improved customer retention to approximately 96% and added roughly 140 revenue-producing trucks sequentially in the second quarter, signs that contracted relationships remain sticky. If the model were structurally impaired, retention and conversion would be deteriorating, not holding.
The risk to that view is duration. A 5% to 10% sequential earnings decline is not trivial if it extends beyond one quarter, and the company's warning suggests management sees cost pressure persisting through the third quarter. Investors will get clarity when J.B. Hunt reports third-quarter results; until then, the stock is pricing the possibility that the cost wave is larger than the demand recovery.
There is a second-order link between the two stories that the market is not pricing. If Washington succeeds in pulling more semiconductor capacity onto U.S. soil — through deals like the one SK Hynix is discussing, or through tariff and subsidy pressure — the industrial fuel and freight bill for moving that equipment, materials, and finished inventory rises with it. Onshoring is a capital-intensity story for chipmakers; for carriers it is a volume story, and the volume has to arrive before the margin does. J.B. Hunt's intermodal strength is the closest thing to a leading indicator that the semiconductor supply chain is actually moving freight, not just announcing factories.
What to Watch
For Intel and SK Hynix, the falsifying signal is straightforward: a signed agreement — a lease, a joint venture, or a foundry supply contract — would confirm the structural shift, while an exclusive or expanded CoWoS commitment between SK Hynix and TSMC, or a Seoul veto under the Industrial Technology Protection Act, would undercut it. A third signal sits between them: if Intel lands a different anchor tenant for Ohio before SK Hynix commits, the memory maker's negotiating leverage weakens and the deal terms shift. For J.B. Hunt, the test is the third-quarter print: if the sequential earnings decline exceeds 10% or management signals that cost pressure will persist into the fourth quarter, the cyclical-recovery thesis is wrong.
Short term, both stocks are trading on headline risk: Intel and SK Hynix on the chance of a deal, J.B. Hunt on the certainty of a cost squeeze. Medium term, the onshoring story has more structural support because it addresses a capacity bottleneck that will not clear on its own and a geopolitical imperative that will not go away. Long term, Intel's foundry turnaround still depends on execution across nodes, not just filling an Ohio campus, while J.B. Hunt's recovery depends on the freight cycle completing its turn.
The market is drawing a clean line between two kinds of inflation: the kind that creates a bottleneck rent for the company that can relieve it, and the kind that compresses the carrier that must absorb it. Intel is being paid for the first; J.B. Hunt is being punished for the second.
Explore more exclusive insights at nextfin.ai.
