SK hynix Supplies 12-Layer HBM4E Samples to Accelerate AI Chip Innovation

High-bandwidth memory is a critical component in AI accelerators, providing the data throughput required by advanced processors. HBM4E is the next step in that product cycle, and its base logic die requires sophisticated foundry manufacturing. Adding a second supplier could help SK Hynix contain rising process costs and reduce dependence on a single manufacturing partner, while challenging TSMC’s reported exclusive position in the company’s outsourced HBM logic-die production.

SK Hynix is evaluating whether to allocate part of its HBM4E base-die production to Intel Foundry, according to the latest report, creating a dual-source arrangement alongside TSMC. As of Aug. 31, 2026, the plan remained under consideration, with no disclosed production schedule, process node, wafer volume or contract value. A deal would give Intel a foothold in the fast-growing AI memory supply chain and could weaken TSMC’s exclusive role in the program.

SK Hynix has become a prime beneficiary of the AI infrastructure boom, using its lead in high-bandwidth memory, or HBM, to supply chips that feed data to AI accelerators. Demand for HBM, server DRAM and enterprise solid-state drives, combined with sharply higher memory prices, has turned a cyclical recovery into a semiconductor supercycle. KB Securities estimated that global technology companies and AI data-center operators represented about 70% of second-quarter revenue.

On July 29, SK Hynix reported 2026 second-quarter revenue of 79.32 trillion won and operating profit of 60.54 trillion won, up 257% and 557% from a year earlier, respectively. Its operating margin reached 76%. Net profit surged to 93.92 trillion won, producing a 118% net margin, as all three measures set quarterly records. Even so, revenue and operating profit missed a 14-brokerage consensus of 84.1 trillion won and 64.1 trillion won, showing how elevated expectations have become.

High-bandwidth memory, or HBM, has become critical to AI chip performance as demand for AI computing surges. South Korean semiconductor giant SK Hynix plays a central role in its partnership with AI chip leader Nvidia. The latest HBM4 technology is both a key benchmark for AI servers and a decisive battleground for memory manufacturers seeking market dominance, making it significant to the global AI supply chain.

SK Hynix has formally begun mass production and shipments of 12-layer HBM4 for Nvidia’s next-generation Vera Rubin AI platform. Simplified validation and emergency production by the two companies have substantially shortened the shipment schedule, with volumes expected to expand significantly from September 2026. Market estimates indicate that SK Hynix could secure about 60%–70% of Nvidia’s HBM4 supply, maintaining its lead over Samsung and Micron in the global commercialization race.

SK Hynix, one of the world’s leading memory suppliers, had planned to expand production of sixth-generation high-bandwidth memory, or HBM4, to meet demand for AI chips. But tightening supply and demand for conventional DRAM and improving margins have prompted the company to reallocate capacity. The move could strengthen profitability, while giving Samsung Electronics an opportunity to narrow SK Hynix’s lead in HBM market share.

As of July 20, 2026, market sources said SK Hynix would delay its HBM4 mass-production and production-line conversion schedules to prioritize shortages of conventional DRAM. The company has not disclosed the length of the delay, the amount of capacity to be shifted or the investment involved. Following the reports, Taiwanese memory stocks Nanya Technology and Winbond Electronics extended their declines in early trading.

AI accelerators must transfer large volumes of data rapidly between processors and memory, making high-bandwidth memory a critical part of the AI chip supply chain. SK hynix is a major HBM supplier, while TSMC has expertise in advanced logic processes and packaging. Combining their capabilities in memory, base dies and packaging could help meet cloud providers’ demand for custom AI chips.

SK Group Chairman Chey Tae-won recently met TSMC Chairman and CEO C.C. Wei in Taipei, where they agreed to expand cooperation on HBM4 and advanced packaging. SK hynix will outsource production of its HBM4 base dies to TSMC to strengthen custom-product development. Available information does not disclose the meeting’s exact date, the value of any investment or contract, or the mass-production timetable.

High-bandwidth memory (HBM) delivers large volumes of data to AI accelerators with high bandwidth and low power consumption, making it a critical component in the race for generative AI computing power. After a downturn in the memory market, SK hynix rebounded through technology investment and its supply partnership with Nvidia, becoming a key company in the global AI semiconductor supply chain.

South Korean broadcaster KBS was granted access to SK hynix’s M16 wafer fabrication plant for the first time. Its program “The World Seen for the First Time” showed the company’s secret HBM production line operating around the clock, along with employees at work. The report did not specify an air date, investment amount or production capacity, but footage from the facility showed the company working at full speed to meet demand for AI memory.

High-bandwidth memory, or HBM, vertically stacks multiple memory dies to deliver greater bandwidth for AI accelerators and has become a critical component of generative AI infrastructure. SK Hynix is now focusing on hybrid bonding, which shortens the connections between chips to increase transfer speeds and improve heat dissipation in stacked designs. The technology is central to its competition with Samsung Electronics and Micron.

SK Hynix said it has completed validation of a 12-layer HBM stack and is expanding its hybrid bonding capabilities, with the next phase targeting commercialization of 16-layer HBM in 2027. Compared with current 12-layer designs, 16-layer products are expected to increase capacity per chip and improve data-processing performance. However, the reports did not disclose a formal launch date, a mass-production timetable or the amount of investment.

High-bandwidth memory, or HBM, is a critical component that gives AI accelerators rapid access to data, and supply has fallen short as data centers expand worldwide. SK hynix currently controls about 60% of the HBM market and is using advanced packaging and next-generation products to strengthen its lead while competing with Samsung Electronics and Micron for orders from major cloud providers.

SK Group Chairman Chey Tae-won recently pledged to sharply increase SK hynix's HBM output and accelerate the latest HBM4 into mass production to narrow the supply-demand gap for AI chips. He said adding wafer capacity takes at least four years and that tight global supplies could persist until 2030. The reports did not disclose the investment amount for the expansion or a specific production start date.