Samsung Electronics and SK Hynix posted a combined 150 trillion won in operating profit for the quarter ended June 30, as AI-server demand and higher memory prices lifted both companies to records. Samsung reported 89.5 trillion won on July 30, while SK Hynix reported 60.5 trillion won a day earlier. Both said supply remains tight and described multiyear customer agreements, extending the AI buildout's effects beyond GPU demand.
Samsung Electronics and SK Hynix posted a combined 150 trillion won in operating profit for the quarter ended June 30, with both companies tying record results to AI-server demand and higher memory prices. Samsung reported its results on July 30, one day after SK Hynix.
Samsung recorded 171.5 trillion won in revenue and 89.5 trillion won in operating profit. Its Device Solutions division, which includes semiconductors, contributed 127.5 trillion won in revenue and 89.2 trillion won in operating profit. Samsung said its memory business set quarterly records after prioritizing server products, expanding HBM4 sales and shipping HBM4E samples to major customers.
SK Hynix reported 79.3 trillion won in revenue and 60.5 trillion won in operating profit, both company records. It attributed the quarter to rising prices and greater sales of high-value products including high-bandwidth memory, AI-server DRAM and enterprise SSDs. The company said it began mass shipments of HBM4 during the quarter.
Long-term agreements move into the foreground
The releases also showed that buyers are looking beyond one quarter. SK Hynix said it had finalized long-term agreements with about 10 customers and was discussing additional deals. The Associated Press separately reported that Samsung had secured long-term supply contracts with five major global data-center customers and was in talks with other technology companies.
Those agreements can improve supply visibility for both buyers and manufacturers. They also raise the cost of forecasting mistakes: memory capacity takes time and capital to expand, while customers must estimate how quickly AI training, inference and data-center demand will grow.
Supply remains the operating constraint
Samsung expects server DRAM, enterprise SSD and HBM demand to accelerate in the second half of 2026 and said the market should remain undersupplied. SK Hynix likewise said customer demand exceeds its current supply capability and outlined phased capacity expansion.
For ML infrastructure teams, the practical signal is that memory availability and price remain first-order inputs to cluster design. HBM constrains how accelerators handle model weights and intermediate data, while server DRAM and storage affect data pipelines, caching and inference systems. The results do not settle whether the current investment cycle will remain durable: AP reported that both companies' shares fell amid concern about heavy capacity spending and stronger Chinese competition. They do show that AI demand is now materially shaping the economics and contracting structure of the memory market.
Key Points #
- 1Samsung reported 89.5 trillion won and SK Hynix 60.5 trillion won in June-quarter operating profit, taking their combined result to about 150 trillion won.
- 2Both companies tied record memory results to AI-server demand, higher prices and high-value products including HBM, server DRAM and enterprise SSDs.
- 3Long-term customer agreements may improve supply visibility, but they make capacity planning and AI-demand forecasts more consequential.
Scoring Rationale #
Official June-quarter releases from Samsung Electronics and SK Hynix show record operating profit and describe AI-server memory demand, supply constraints and long-term customer agreements. The event is materially relevant to AI infrastructure planning because HBM, server DRAM and enterprise SSD availability affect accelerator deployment, data pipelines and procurement, while the article preserves the unresolved risk around capacity spending and demand durability.
Sources #
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.