Oct 27, 2026 earnings call — Shareholder returns and HBM4 ramp
Announced a shareholder return plan of approximately KRW 100 trillion, including cash dividends of KRW 30 trillion or more.
Revenue $78.2B (est $71.4B) · EPS $4.01 (est $6.20)
Key quotes
"We plan to execute cash dividends of KRW 30 trillion or more and share acquisition/cancellations of KRW 60 trillion or more, including the KRW 40 trillion announced in August."
"In particular, full-scale mass production and shipments of HBM4 commenced in the third quarter, additionally, shipments of server DDR5 increased by over 40% compared to Q2, driving overall growth in DRAM bit shipments."
"The volume secured under LTAs currently accounts for more than 50% of our total capacity for 2027 and 2028, establishing a foundation for mid- to long-term profit stability and demand visibility."
Analyst Q&A 8
Why AI investment will keep growing
Q (Sei Cheol Lee · Citi): The analyst asked why AI investment momentum will continue, given growing AI capex but slow monetization and market concern.
A: Management said customers are reexamining project timing but gave 2 main reasons to expect continued AI investment. First, customers' AI projects are already driving adoption and business results, with major hyperscaler cloud revenue up 30% to 80% Y-o-Y and cloud order backlogs growing twofold to fivefold Y-o-Y. Second, the investor base is diversifying beyond big tech to include nontech companies and sovereign AI, so cuts by some can be offset by others; memory supply-demand imbalance where demand outpaces supply is expected to continue.
"There are 2 main reasons behind our view that the AI investment momentum will continue." — Unknown Executive
Why dividend vs buyback split was chosen
Q (Sunwoo Kim · Meritz): The analyst asked how the company decided the split between cash dividend and share buybacks in the shareholder return plan.
A: Management said they balanced direct cash returns and raising per-share value through buybacks. They set the cash dividend at a sustainable level reflecting performance, historical dividend yield, global peers' yield and appropriate cash holdings, and will pay most investment returns (mainly from Kioxia) as special dividends partly early in the third quarter and the rest at year-end; after cash dividends the remaining total shareholder return goes to share acquisition, executed each quarter within the annual size and set by the Board.
"First, we decided on cash dividend at a level that will be sustainable." — Unknown Executive
Long-term agreements (LTA) and sales visibility
Q (Giuni Lee · Goldman Sachs): The analyst asked what 50% LTA coverage of capacity for 2027-2028 means for performance and long-term growth potential.
A: Management said raising LTA coverage improves longer-term demand visibility and supports stable revenue and earnings growth. They said DRAM and HBM volumes through 2029 are effectively sold out when combining volumes already secured under LTAs and volumes under discussion, that LTA volumes account for more than 50% of total capacity in '27 and '28 and will stay similar, and that non-LTA capacity will be used flexibly for market conditions and next-generation products.
"By increasing LTA coverage, our goal is to improve longer-term demand visibility, which can lead to stable revenue and earnings growth." — Unknown Executive
How company will avoid overproduction
Q (Simon Woo · BofA): The analyst asked what the company would do if demand falls short of expectations given big fab and capacity investments.
A: Management said current investment builds medium- to long-term infrastructure for tech migration and high value-add products, not simply to raise bit supply, and that investment will be staged. The foundational shell will be prepared in advance but equipment move-in and mass production ramps will be adjusted to demand; they also review customers' financial health and maintain financial room to manage scale and pace of investment, and will flexibly adjust disbursement, production plans and resource allocation if conditions differ.
"The foundational infrastructure will be ready in advance for medium- to long-term demand, but the key resources needed to increase supply, such as equipment move-in and mass production ramp-up will be adjusted in line with demand." — Unknown Executive
HBM4 supply, yields, and next products
Q (H. Kwon · JPMorgan): The analyst asked for specifics on HBM4 supply and plans for HBM4E and custom HBM.
A: Management said HBM4 mass production is proceeding without disruption and is expected to cross over HBM3E in the fourth quarter in both bit shipment and revenue share, with stable, industry-best yields; HBM4 already exceeded planned yield and HBM3E surpassed planned mature yield. They are maximizing productivity to meet demand, have completed HBM4E sample shipment in the second quarter with development and qualification on customers' timelines, and are working with major customers on growing custom HBM projects.
"HBM4 is expected to cross over HBM3E in the fourth quarter in both bit shipment and revenue share, thanks to mass production proceeding without disruption." — Unknown Executive
Impact of lower HBM stack counts
Q (Dong Hee Han · SK): The analyst asked about the possibility and impact of customers moving to shorter HBM stacks on shipments and profitability.
A: Management said customers lowering stack counts is a response to constrained HBM supply to get more cubes, but lower stacks do not necessarily mean a proportional capacity drop because increased capacity per die can offset much of the impact. They said medium- to long-term HBM demand outlook remains unchanged, supply plans already account for more cubes and back-end needs, and profitability depends on product mix, yield and costs; they expect HBM revenue and profit to grow significantly Y-o-Y next year with improved profitability and to remain high.
"So even when considering such adjustments, our medium- to long-term HBM demand outlook remains unchanged." — Unknown Executive
Progress on 4F square DRAM technology
Q (Young-gun Kim · Mirae Asset): The analyst asked for an update on 4F square preparation and areas of development to keep technology competitiveness.
A: Management said they will extend 6F square through process maturity and EUV-based scaling, then transition to 4F square or 3D DRAM when ready; applicability depends on scaling capability, process environment and cost. They are preparing a step-by-step roadmap from extreme scaling of existing DRAM to 4F square and 3D DRAM and will apply each technology at the right technological and economic timing using process, design and production experience.
"We are preparing a step-by-step road map covering from the extreme scaling of the existing DRAM to 4F square and 3D DRAM." — Unknown Executive
NAND demand from KV cache and memory tiering
Q (Young Ho Ryu · NH Investment &): The analyst asked how growing KV cache and tiered memory will affect NAND demand and what preparations the company is making.
A: Management said KV cache is growing and tiered memory across HBM, DRAM and SSD is becoming important because HBM/DRAM cannot hold all KV cache. They are developing memory tiering technologies such as SALT-KV and CMM-hybrid, running system demonstrations for eSSDs in real inference environments, and plan to address a new memory tier between HBM and SSD with HBF to capture the new NAND demand opportunity.
"And in response to this opportunity, the company is developing memory tiering technologies including the sematic-aware life cycle tiering for KV and CMM-hybrid." — Unknown Executive