Cantor Fitzgerald Says SK Hynix’s U.S. Shares Could Nearly Double Within a Year.
SK Hynix has barely settled into its new home on the Nasdaq, but one Wall Street firm already believes the South Korean chipmaker could deliver one of the market’s most striking gains.
Cantor Fitzgerald began coverage of SK Hynix’s American depositary receipts with an “overweight” rating and a $300 price target. That forecast implies the U.S.-listed shares could nearly double over the next 12 months, driven by rising artificial intelligence spending, limited memory-chip supplies and the company’s leading position in high-bandwidth memory.
The bullish call immediately attracted attention. SK Hynix’s U.S. shares climbed to about $155 during Tuesday afternoon trading, gaining more than 8% and reaching an intraday high of $155.46. Cantor’s $300 target would still represent approximately 94% upside from that level.
It is an ambitious forecast, but Cantor is not alone. At least six brokerages launched positive coverage of SK Hynix on Tuesday, with Rosenblatt Securities setting an even higher target of $320. The sudden wave of enthusiasm reflects a growing belief that memory chips are becoming as important to the AI economy as the processors receiving most of the public attention.
Wall Street Sees More Than Another Chip Cycle
Memory-chip manufacturers have traditionally operated in a punishing boom-and-bust industry. Strong demand encourages companies to increase production, excess supply eventually pushes prices down, and profits can disappear almost as quickly as they arrived.
Cantor Fitzgerald believes the current AI expansion could weaken that familiar pattern.
The firm’s analysts expect demand for DRAM and NAND memory to exceed available supply through 2029, potentially giving manufacturers stronger pricing power and more predictable earnings. Generative AI, autonomous software agents, data centers and physical AI systems such as robots will all require larger amounts of fast, energy-efficient memory.
That shift matters because processors cannot handle increasingly complex AI models alone. They need rapid access to enormous volumes of data. When memory cannot deliver that information quickly enough, even the most advanced processor becomes less effective.
High-bandwidth memory, commonly known as HBM, was designed to solve that bottleneck. It stacks memory components vertically and places them close to the processor, allowing data to travel faster while using less power than many conventional memory systems.
SK Hynix has built a strong position in this market. The company supplies advanced memory used in AI accelerators and works closely with some of the technology industry’s largest customers.
That leadership is one reason Bank of America described SK Hynix as undervalued. The bank pointed to strong orders from U.S. technology companies, the chipmaker’s position in premium memory and the possibility of a long-lasting earnings “supercycle” as AI infrastructure spending increases.
The $300 Target Rests on an Expanding AI Economy
Cantor’s forecast is ultimately a bet that artificial intelligence infrastructure spending will remain elevated for years rather than quarters.
Technology companies are racing to build larger data centers filled with advanced processors, networking equipment and storage systems. Every new AI facility requires substantial quantities of memory to train models, store information and deliver responses quickly.
That demand is pushing memory away from its old image as a replaceable commodity. Advanced HBM products are difficult to manufacture, require specialized packaging and must meet strict performance and power-efficiency standards.
SK Hynix strengthened its position in June by announcing a multi-year technology partnership with Nvidia. The companies plan to co-develop next-generation memory aligned with Nvidia’s future AI infrastructure, personal computing and robotics platforms.
The agreement covers memory for Nvidia’s Vera Rubin AI supercomputers, Vera central processing units, RTX Spark-powered computers and Jetson Thor robotics systems. SK Hynix and Nvidia also intend to use AI tools to improve semiconductor design and manufacturing.
The partnership gives SK Hynix something more valuable than a large customer. It places the company closer to Nvidia’s product-development roadmap, allowing both businesses to coordinate before future computing platforms reach the market.
That cooperation could help SK Hynix prepare its manufacturing capacity earlier and design products around the exact requirements of upcoming AI systems. It could also make it more difficult for competitors to displace the company after its memory has been integrated into a customer’s broader architecture.
SK Hynix has also discussed deeper cooperation with Meta. The companies have explored long-term HBM supplies for Meta’s custom AI accelerators, future products beyond HBM4, and possible memory solutions for AI-powered wearable devices.
These relationships show why Wall Street increasingly views the company as an AI infrastructure partner rather than simply another memory supplier.
Record Results Provide Support for the Bullish Call
Cantor Fitzgerald’s optimism is not based entirely on future expectations. SK Hynix is already reporting extraordinary growth as demand for premium AI memory rises.
The company said second-quarter revenue increased 257% from the same period a year earlier, while operating profit jumped 557%. It described the quarter as its strongest on record, supported by higher prices and increased sales of HBM, AI-server DRAM and enterprise solid-state drives.
SK Hynix also reported that first-half revenue exceeded 100 trillion won for the first time in its history. Its cash position strengthened, while total debt declined during the quarter.
The company has completed long-term supply arrangements with around 10 major customers and continues negotiating with others. Multi-year contracts could reduce some of the uncertainty that has traditionally made memory manufacturers so vulnerable to sudden demand changes.
SK Hynix began mass shipments of HBM4 during the second quarter and plans to expand production during the second half of 2026. The company said the product meets customer speed requirements while offering competitive power efficiency and production costs.
Those details are important because the AI race is not only about creating faster chips. Data-center operators must also control electricity consumption, cooling needs and operating expenses. Memory that moves information quickly while consuming less power can improve the economics of an entire computing system.
The company has also started mass production of a 192-gigabyte SOCAMM2 module developed for Nvidia’s Vera Rubin platform. The technology adapts low-power memory commonly associated with mobile devices for use in next-generation AI servers.
Together, these products give SK Hynix exposure to multiple parts of the AI memory market rather than a single generation of HBM.
The Nasdaq Listing Could Unlock a Higher Valuation
SK Hynix began trading its depositary receipts on the Nasdaq on July 10, giving U.S. investors a direct way to own the company through an American exchange.
The listing was designed to broaden the chipmaker’s investor base and strengthen its profile at the center of the global AI market. The company raised approximately $26.5 billion after pricing the depositary receipts at $149, supported by intense interest in AI-focused semiconductor businesses.
The shares initially struggled as semiconductor stocks pulled back, but analysts believe the Nasdaq listing could eventually narrow the valuation gap between SK Hynix and U.S.-based competitor Micron Technology.
William Blair said SK Hynix could move closer to Micron’s valuation as American investors become more familiar with the company. Its analysts argued that the listing, stronger long-term earnings visibility and close connection to AI data-center spending could support a structural revaluation.
That revaluation is a major part of the road to $300.
Cantor is not merely predicting that SK Hynix will sell more chips. Its target suggests investors may become willing to pay a higher multiple for the company’s profits because AI demand could make those earnings more durable than they were during previous memory cycles.
William Blair set a $260 target, while Stifel assigned a $240 target. Rosenblatt’s $320 forecast was the most optimistic among the new analyst calls.
The targets differ, but the broader message is consistent: several Wall Street firms believe the market has not yet fully valued SK Hynix’s technological lead and earnings potential.
SK Hynix Is Preparing for Demand Beyond HBM
The company is already developing technologies intended to address the next major AI memory challenge.
SK Hynix and Sandisk are working to standardize high-bandwidth flash, or HBF. The proposed technology aims to combine much of the capacity offered by NAND flash storage with speeds closer to high-bandwidth memory.
HBF could eventually fill the gap between extremely fast HBM and larger but slower storage systems. That may become increasingly important as AI inference systems process larger datasets and require fast access to huge libraries of information.
The company is also expanding production. SK Hynix plans to invest approximately $3.87 billion in an advanced packaging and research facility in West Lafayette, Indiana. The site is expected to begin mass production during the second half of 2028 and create roughly 1,000 jobs.
Advanced packaging is critical because HBM requires multiple memory layers to be assembled with precision. Increasing production capacity in the United States could bring SK Hynix closer to important customers while strengthening its supply chain.
A Near-Doubling Still Comes With Serious Risks
Cantor Fitzgerald’s $300 target is a forecast, not a guarantee.
SK Hynix operates in a highly competitive and capital-intensive industry. Samsung Electronics and Micron are investing heavily in advanced memory, while Chinese manufacturers are attempting to close the technology gap.
There is also a risk that the industry expands production too aggressively. If new factories begin supplying more memory than the market needs, prices could fall and recreate the oversupply cycles that have repeatedly hurt chipmakers.
Recent earnings also delivered a warning. Although SK Hynix reported record profits, its results fell short of some analyst forecasts because of delays in shipments of advanced memory products. The disappointment contributed to concerns about how quickly AI investments will translate into completed orders.
The U.S. shares have already shown considerable volatility since their July listing. They have traded between approximately $124.80 and $194.80, demonstrating how quickly sentiment can change around companies tied to the AI boom.
Still, Cantor’s call captures an important change in the semiconductor industry. Memory is no longer a quiet component operating behind the processors. It has become one of the most valuable and difficult-to-replace parts of the AI computing system.
SK Hynix does not need every bullish prediction to come true for its U.S. shares to rise. It needs artificial intelligence spending to remain strong, advanced memory supplies to stay tight, and its technological lead to survive increasing competition.
Cantor Fitzgerald believes those conditions could push the shares to $300 within 12 months. Reaching that target would require SK Hynix to execute almost flawlessly. Still, Wall Street’s message is becoming difficult to ignore: the next major winner of the AI boom may be the company supplying the memory that allows everything else to work.
