Semiconductor stocks have started the year on a strong footing, with gains concentrated not in flashy logic chips but in a quieter, more fundamental corner of the industry: memory.
Shares of the world’s biggest memory makers have surged, reflecting how artificial intelligence is reshaping demand patterns across the chip sector and tightening supply in critical components.
South Korea’s SK Hynix and Samsung Electronics, the two largest memory chipmakers globally, are up 11.5% and 15.9% respectively so far this year. In the United States, Micron has climbed 16.3%. The rally comes as investors bet that AI-related demand, which drove chip markets through 2025, is not fading but intensifying.
Register for Tekedia Mini-MBA edition 19 (Feb 9 – May 2, 2026): big discounts for early bird.
Tekedia AI in Business Masterclass opens registrations.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab: From Technical Design to Deployment (next edition begins Jan 24 2026).
At the heart of the move is memory’s central role in AI computing. Training and running large AI models designed by companies such as Nvidia and AMD requires vast amounts of fast, high-capacity memory to move data efficiently between processors. As cloud providers and tech giants pour billions of dollars into AI data centers, memory has emerged as a bottleneck.
One segment has been especially important: dynamic random-access memory, or DRAM, used extensively in AI servers. Prices for DRAM surged sharply in 2025 as demand outpaced supply, and that pressure has not eased. Counterpoint Research expects memory prices to rise another 40% through the second quarter of 2026, extending what analysts increasingly describe as a full-blown cycle rather than a brief spike.
“The recent rally across the semiconductor space has been driven largely by the memory side of the market rather than logic chips,” Ben Barringer, head of technology research at Quilter Cheviot, said in an email to CNBC. “We’re seeing a combination of very strong demand from AI workloads and relatively constrained supply, particularly in high-bandwidth memory, which is essential for training and running large AI models.”
High-bandwidth memory, or HBM, has become one of the most sought-after components in the AI supply chain. It sits close to processors in advanced packaging configurations, enabling faster data transfer and lower power consumption. SK Hynix is widely seen as a leader in this area, supplying HBM used in some of Nvidia’s most powerful AI accelerators, a position that has strengthened its earnings outlook.
That backdrop explains the optimism heading into earnings season. Samsung is expected to report a 140% jump in fourth-quarter operating profit, according to LSEG estimates, marking a sharp turnaround after a prolonged downturn in its memory business. Micron’s earnings per share are forecast to rise more than 400% year-on-year in the December quarter, reflecting both higher prices and improving utilization rates.
The rally has spilled beyond memory producers themselves. Investors are increasingly positioning for a broader AI-driven expansion across the semiconductor value chain. Intel shares are up 7.6% year-to-date, while Taiwan Semiconductor Manufacturing Co., the world’s largest contract chipmaker, has gained 10%. Both companies manufacture a wide range of chips and are expected to benefit as customers ramp up spending on advanced semiconductors tied to AI workloads.
Equipment suppliers are also riding the wave. ASML, the Dutch firm whose lithography machines are essential for producing the most advanced chips, has seen its shares rise 15.2% this year. Bernstein on Sunday raised its price target on ASML from 800 euros to 1,300 euros, implying about 24% upside from Tuesday’s trading level.
“ASML stands to benefit enormously from the wave of capacity expansion planned for 2026 and 2027,” Bernstein analysts wrote, pointing specifically to memory. They said the company would gain “from the upcoming DRAM super cycle,” as manufacturers invest heavily in new fabs and more advanced production lines.
That link is crucial. As memory makers respond to tight supply and strong pricing by expanding capacity, demand for ASML’s tools rises in tandem. Advanced DRAM and HBM production requires cutting-edge manufacturing equipment, locking ASML deeper into the AI investment cycle.
Recent signals from industry executives have reinforced the bullish narrative. SK Hynix has pointed to the possibility of an extended HBM supercycle, suggesting demand could remain elevated well beyond a single year.
“Recent comments from SK Hynix pointing to a potential HBM supercycle have reinforced the idea that this is not just a short-term bounce, but a more structural shift linked to the ongoing build-out of AI infrastructure,” Barringer said. “That has helped improve sentiment across the sector, especially for companies with direct exposure to AI-driven memory demand.”
The emerging picture is one where memory, long treated as the most cyclical and volatile part of the semiconductor industry, has become central to the AI story. As long as companies continue to scale data centers and push larger, more data-hungry models, memory demand is likely to stay tight.
For investors, this has reframed how the semiconductor rally is being judged. This is not simply about who designs the smartest AI chips, but about who controls the components that make those chips usable at scale. So far in 2026, memory makers are winning that argument.