Samsung Electronics Surpasses 30% NAND Market Share, Bolsters 'Super-Gap' Lead Through NVIDIA Alliance

[Korea Financial Times, Gwak Horyung] Samsung Electronics (Co-CEOs Jeon Young-hyun and Roh Tae-moon) has leveraged the generative artificial intelligence (AI) boom to more than double its NAND flash revenue quarter-over-quarter in the first quarter of 2026, further cementing its dominant position as the world's No. 1 chipmaker in the segment. With the NAND flash market also experiencing a historic super-cycle following record-breaking demand in DRAM, Samsung plans to accelerate its push to capture the high-performance AI storage market through the launch of a new product in the second half of the year — one confirmed for integration into NVIDIA platforms.

According to market research firm TrendForce, Samsung Electronics recorded NAND flash revenue of USD 13.51 billion (approximately KRW 20 trillion) in Q1 2026. This represents a 104.7% increase from the prior quarter and a more-than-threefold rise compared to the same period last year — the highest growth rate among the top five NAND flash vendors. Its market share also climbed from 28% in the prior quarter to 31.6%.

During the same period, combined revenue of the top five NAND flash vendors rose 83.7% to USD 38.9 billion. SK hynix ranked second with USD 7.5 billion (17.6% market share). Japan's Kioxia and U.S.-based Micron and SanDisk occupied the third through fifth positions, each recording a revenue share of 13.9%.

The rapid expansion of the NAND flash market is being driven primarily by the proliferation of generative AI services. NAND flash is a type of non-volatile memory semiconductor that retains data even when power is cut off. As AI inference processes generate exponentially increasing volumes of KV (key-value) cache data — used to maintain preceding context and long-term memory — data centers are actively investing in high-performance NAND flash-based storage infrastructure.

During its Q1 2026 earnings conference call, Samsung Electronics noted that server-use NAND flash bit growth (sales volume growth) for the quarter came in at the low-20% range compared to the prior quarter, surpassing initial forecasts.

NAND Operating Margin Eyes 70% Range as Segment Gains Prominence

Samsung's Device Solutions (DS) division — which encompasses its semiconductor operations — posted revenue of KRW 81.7 trillion in Q1. NAND flash accounted for approximately 15% of that figure. DRAM, which has led the AI-driven boom, remained the largest contributor at approximately 65% of DS revenue, while non-memory segments including System LSI and foundry accounted for roughly 9%.

Meritz Securities estimates that Samsung's NAND flash division posted an operating profit margin of 63.9% in Q1, a sharp improvement from 24.8% in the preceding quarter. The DRAM operating profit margin stood at approximately 78%. The brokerage further projected that Samsung's NAND flash revenue will reach KRW 33 trillion in Q4 2026, with an operating profit margin in the mid-to-upper 70% range.

Across the semiconductor industry, the prevailing view is that NAND flash selling prices will remain elevated through year-end, as enterprise demand — particularly from server customers — continues to expand while major NAND flash vendors' capacity additions remain constrained.

Samsung NAND Earmarked for NVIDIA's 'Vera Rubin'

Looking ahead, Samsung Electronics plans to launch new PCIe Gen 6 enterprise NAND flash storage products based on V9 TLC (9th-generation Vertical NAND, Triple-Level Cell) architecture in the second half of this year, with the aim of extending its "super-gap" lead as the global No. 1 player.

The product in question is the PM1763, which Samsung unveiled at NVIDIA's GTC conference in March. Samsung has confirmed that the PM1763 will serve as the primary storage solution in NVIDIA's next-generation AI platform, "Vera Rubin."

Compared to the previous fifth-generation product, the PM1763 delivers up to twice the data transfer bandwidth and 1.8 times greater power efficiency. Its form factor has also been slimmed down to accommodate the liquid cooling designs increasingly adopted by enterprise customers. The device will support storage capacities ranging from 4TB up to a maximum of 64TB.

"NVIDIA has proposed a storage architecture in which data generated during AI inference is stored not only in High Bandwidth Memory (HBM) but also in NAND-based storage," Samsung Electronics said. "We will take the lead in the market by capturing early demand from AI servers and data centers."

Gwak Horyung (horr@fntimes.com)