🔥36氪•Stalecollected in 1m
Samsung and SK Hynix drive Korean profit surge
💡See how AI-driven memory demand is reshaping the global semiconductor financial landscape.
⚡ 30-Second TL;DR
What Changed
Q1 operating profit for top 500 Korean firms reached 156.35 trillion KRW.
Why It Matters
The dominance of memory chip manufacturers underscores the critical bottleneck and value concentration in the AI hardware supply chain.
What To Do Next
Analyze HBM (High Bandwidth Memory) production capacity trends to forecast potential hardware supply constraints for your AI infrastructure projects.
Who should care:Founders & Product Leaders
Key Points
- •Q1 operating profit for top 500 Korean firms reached 156.35 trillion KRW.
- •Samsung and SK Hynix account for over 60% of the total profit.
- •The surge is largely attributed to the high demand for AI-related memory chips.
🧠 Deep Insight
Web-grounded analysis with 22 cited sources.
🔑 Enhanced Key Takeaways
- •South Korea's economy experienced a stronger-than-expected 1.7% GDP growth in the first quarter of 2026, marking its fastest quarterly expansion in five and a half years, primarily driven by robust semiconductor exports.
- •The South Korean AI market is projected to reach USD 4.34 billion in 2025 and is anticipated to grow at a compound annual growth rate (CAGR) of 27.63% from 2025 to 2030, underscoring the nation's strategic focus on AI.
- •The intense demand for AI-related memory chips is leading to 'memflation,' with Gartner forecasting DRAM and NAND flash annual prices to increase by 125% and 234% respectively in 2026, and meaningful pricing relief not expected until late 2027.
- •Data centers are projected to consume up to 70% of all high-end memory in 2026, a significant shift from previous years, as memory manufacturers reallocate production capacity towards high-margin High Bandwidth Memory (HBM).
- •While SK Hynix held a dominant 62% share of the HBM market in Q2 2025, with Samsung at 17% and Micron at 21%, Samsung is projected to surpass a 30% HBM market share in 2026, driven by its HBM4 products entering NVIDIA's supply chain.
📊 Competitor Analysis▸ Show
| Feature/Metric | SK Hynix | Samsung | Micron Technology |
|---|---|---|---|
| HBM Market Share (Q2 2025) | 62% | 17% | 21% |
| HBM3E Status | Market leader, secured early supply agreements with NVIDIA for H100/H200 series | HBM3E validation completed in 2025, mass production ramp | Shipping 12-stack HBM3E in 8-high and 12-high configurations |
| HBM4 Status | Under development with key customers, will be integrated into NVIDIA's Rubin platform | Began mass production shipments of 12-layer HBM4 in Q1 2026, uses 1c DRAM and 4nm base die, achieving up to 13 Gbps | Samples with key partners, offers 11 Gbps data transfer speeds and total bandwidth above 2.8 TB/s |
| Projected HBM Market Share (2026) | 54% (HBM4) | >30% (overall HBM), 28% (HBM4) | Remains a key player, no specific 2026 projection found, but shipping HBM3E and sampling HBM4 |
🛠️ Technical Deep Dive
- High Bandwidth Memory (HBM) is a computer memory interface for 3D-stacked synchronous dynamic random-access memory (SDRAM).
- HBM achieves high memory bandwidth by stacking multiple DRAM dies vertically and connecting them with Through-Silicon Vias (TSVs) and microbumps, often packaged alongside the processor die using 2.5D interposers.
- Key HBM generations include HBM, HBM2, HBM2E, HBM3, HBM3E, and HBM4, with each iteration bringing improvements in processing speed and capacity.
- HBM3E, the fifth generation, delivers over 1.2 TB/s per stack through a 1024-bit interface and 16 independent channels, with per-pin data rates exceeding 9 Gbps.
- HBM4, the sixth generation, doubles the interface width to 2048 bits and 32 independent channels, targeting over 2.0 TB/s per stack, with production expected in 2026.
- Samsung's HBM4 utilizes its advanced 1c DRAM process (10-nanometer class 6th generation) and a 4-nanometer foundry process for the base die, enhancing performance and power efficiency to achieve operating speeds up to 13 Gbps.
- SK Hynix's HBM3E is a standard for NVIDIA's Blackwell GPU, and its HBM4 is slated for integration into NVIDIA's Rubin platform.
🔮 Future ImplicationsAI analysis grounded in cited sources
South Korea's economic growth will remain heavily dependent on the global AI semiconductor cycle.
The country's Q1 2026 GDP growth was primarily driven by semiconductor exports, and its economy is increasingly reliant on its chipmaking firms, making it vulnerable to shifts in overseas demand.
The 'memflation' trend will continue through 2026, impacting the broader electronics market.
Memory manufacturers are prioritizing high-margin HBM for AI, leading to supply constraints and significant price increases for conventional DRAM and NAND, with meaningful relief not expected until late 2027.
Competition in the HBM market will intensify significantly with the ramp-up of HBM4 production.
While SK Hynix currently leads, Samsung is aggressively investing and shipping HBM4, aiming to regain market share, and Micron is also a key player, indicating a fierce battle for dominance in the next-generation AI memory.
⏳ Timeline
2013-10
HBM adopted by JEDEC as an industry standard, with SK Hynix producing the first HBM chip.
2016-01
JEDEC accepted the second generation, HBM2, as an industry standard.
2022-01
JEDEC officially announced the HBM3 standard.
2024-03
SK Hynix started mass production of the world's first HBM3E DRAM.
2025-04
JEDEC officially announced the HBM4 standard.
2026-Q1
Samsung began mass production shipments of 12-layer HBM4 products.
📎 Sources (22)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
- Google Search Source
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 36氪 ↗