SK Hynix plans to strengthen its position in the global memory market by expanding production and research for next generation DRAM technologies as AI continues to increase demand for high performance memory.
Speaking at the 2026 Global Forum, CEO Kwak No Jung said memory has become a fundamental part of modern AI infrastructure and argued that the industry cannot function without it.
The company now plans to use new manufacturing sites and research facilities in South Korea and the United States to support future products including advanced HBM, DDR and LPDDR memory.
AI is increasing the importance of memory
AI systems depend on large amounts of memory to move data between processors and keep accelerators supplied with information.
As AI models become larger, memory bandwidth and capacity can become major performance limits even when processors themselves are powerful.
That has made technologies such as High Bandwidth Memory increasingly important for AI accelerators.
SK Hynix believes this trend will continue.
| Area | SK Hynix plan |
|---|---|
| HBM | Continue development beyond HBM4 |
| HBM4E | Next generation product under development |
| DDR | Prepare future DDR6 technology |
| LPDDR | Advance toward LPDDR6 |
| South Korea production | Expand Yongin semiconductor cluster |
| US production | Develop Indiana manufacturing facility |
| US research | Establish AI focused R&D center |
The company expects demand for AI memory to remain strong as data center operators deploy larger numbers of accelerators.
HBM remains central to SK Hynix's strategy
SK Hynix has become one of the most important suppliers of HBM for AI hardware.
Its latest HBM4 products are already being used with Nvidia's Rubin generation AI accelerators.
The company is also working on HBM4E, which is intended to push memory performance further for future AI systems.
HBM differs from conventional desktop memory because multiple memory dies are stacked vertically and connected using high speed interfaces.
That architecture provides much greater bandwidth close to the processor, making it especially useful for GPUs and AI accelerators that need to process enormous amounts of data.
SK Hynix sees its earlier success with HBM as a model for how it can maintain its technological advantage in future memory generations.
New factories will support future production
One of the company's major projects is its semiconductor cluster in Yongin, Gyeonggi Province in South Korea.
The site is expected to become an important production base for future memory products.
SK Hynix is also building manufacturing capacity in Indiana in the United States.
The company says its international facilities will help support next generation memory production while strengthening its global supply network.
In addition to manufacturing, SK Hynix plans to establish a new AI research and development center in the United States.
That facility will focus on technologies required for future AI memory systems.
DDR6 and LPDDR6 are also in development
AI data centers are only one part of SK Hynix's roadmap.
The company is also preparing future memory standards aimed at PCs, smartphones and other consumer electronics.
LPDDR6 is expected to serve upcoming mobile devices and efficient computing platforms.
DDR6 development is also underway as the industry prepares the successor to DDR5.
These technologies are expected to offer higher bandwidth and other improvements needed by future processors and increasingly demanding software.
That means SK Hynix's next generation strategy covers both data center memory and mainstream computing products.
Competition in memory is becoming stronger
SK Hynix remains one of the three major global memory manufacturers alongside Samsung and Micron.
However, competition is increasing.
Chinese manufacturer CXMT has been expanding rapidly as demand for DRAM grows.
That puts additional pressure on established suppliers to improve manufacturing efficiency, develop more advanced products and secure long term relationships with major customers.
For SK Hynix, maintaining leadership therefore depends on more than simply increasing production capacity.
The company is also investing in memory design, manufacturing processes, packaging and system level technology.
AI demand is reshaping the memory market
The rapid growth of AI infrastructure has changed the role memory plays in the semiconductor industry.
Memory was traditionally treated as a relatively standardized component compared with processors and GPUs.
HBM has changed that dynamic because memory bandwidth is now directly connected to the performance of advanced AI systems.
As accelerators become more powerful, they require increasingly sophisticated memory systems to avoid becoming limited by data movement.
That creates opportunities for companies able to deliver higher bandwidth, larger capacities and more advanced packaging.
SK Hynix is betting that this demand will remain strong enough to justify major investments in new factories and research facilities.
Its strategy now extends from HBM4 and HBM4E for AI accelerators to DDR6 and LPDDR6 for future PCs and mobile devices.
The company expects these technologies, supported by new global production facilities, to help it maintain its position as memory becomes an increasingly important part of the AI computing industry.
Discussion (0)
Be the first to comment.