SK hynix said it expects AI infrastructure investment to continue in the mid to long term despite some big tech corporations reviewing data center leases and the emergence of high-efficiency artificial intelligence (AI) models.
At its second-quarter 2026 earnings conference call on the 29th, SK hynix said, "We are aware of concerns that AI infrastructure investment may be slowing due to some big tech corporations reviewing data center leasing businesses and the rise of high-efficiency AI models," adding, "We view these moves not as a process of cutting back AI investment, but as a process of increasing utilization and accelerating monetization of AI infrastructure built at scale."
It added, "For major cloud service providers (CSPs), AI investment is directly tied to competitiveness in core businesses such as search and advertising, cloud, and software," and "Investment to secure AI competitiveness is likely to remain solid."
It also said the spread of high-efficiency AI models is unlikely to lead to reduced demand for infrastructure and memory. SK hynix said, "When the efficiency of models and systems improves, more users can access diverse services on the same infrastructure, broadening the accessibility and scope of AI services," adding, "Given the explosive user demand even for recent high-efficiency AI models, efficiency gains are working to spur service proliferation and increase overall usage."
SK hynix said that in mid- to long-term demand talks underway with major customers, it is also confirming the sustainability of AI investment. While the timing of individual projects may vary due to physical constraints such as securing power and building data centers, it believes AI infrastructure investment will remain solid beyond next year as AI competition among CSPs and service expansion continue.
SK hynix added, "Accordingly, demand across memory will expand together—from high bandwidth memory (HBM), which directly handles AI computation, to server DRAM for agentic AI, and to high-performance, high-capacity NAND that processes the increase in data driven by the spread of AI services."