AI storage demand surges again: Nvidia CMX may consume over 100 million TB of NAND next year, Samsung has started supplying

```

The demand for NAND flash memory from NVIDIA AI servers is entering a phase of explosive growth. Samsung Electronics, leveraging its production capacity advantage, is quickly turning this trend into a strategic opportunity.

According to a July 20 report from Seoul Economic Daily, Samsung Electronics has started mass production of its tenth-generation V-NAND (V10) and is officially supplying it to NVIDIA. At the same time, about 60% of Samsung's total V-NAND production capacity is allocated to the ninth-generation product (V9) production line to meet the demand for NVIDIA's upcoming next-generation storage device "Context Memory Storage" (CMX). The industry expects the NAND capacity demanded by CMX will soar from 35 million TB this year to over 100 million TB next year.

This wave of demand has a profound impact on the supply structure of the entire NAND market. NVIDIA's large-scale locking-in of NAND means that the supply of storage chips available to other companies and consumers will shrink significantly, placing upward pressure on market prices.

Meanwhile, Samsung Electronics Chairman Lee Jae-yong is expected to meet NVIDIA CEO Jensen Huang in San Francisco soon. The two sides will conduct in-depth discussions on HBM and next-generation NAND product supply, as well as cooperation on building data centers in South Korea.

CMX drives NAND demand surge, NVIDIA locks in Samsung capacity

The birth of NVIDIA CMX stems from the rapid expansion of "Key-Value Cache" (KV Cache) data in AI inference processes. Previously, GPU servers could handle KV cache data internally, but as the scale of data continues to grow, independent external storage devices equipped with thousands of TB-level SSDs have become a necessary configuration for AI servers.

According to wccftech, a single NVIDIA CMX device is equipped with 576 SSDs, with total storage capacity reaching 9,600 TB. The industry estimates CMX's NAND demand this year to be about 35 million TB, and next year to exceed 100 million TB. Independent semiconductor analyst @jukan05 posted on X that the scale of 100 million TB is roughly equivalent to adding a demand source with the size of Apple Inc. to the NAND market, and the impact on the supply side should not be underestimated.

Samsung Electronics currently has a monthly V-NAND wafer production capacity exceeding 100,000. According to reports, before CMX shipment, NVIDIA asked Samsung to expand its NAND capacity, and Samsung responded actively. Samsung’s total NAND production this year is expected to reach around 250 million TB, and the company plans to use this globally leading capacity scale to establish its position as NVIDIA CMX's core supplier.

V9 stable production, V10 mass production, V11 trial production, Samsung advances three generations in parallel

In terms of product line layout, Samsung is simultaneously advancing the production process of three generations of V-NAND products.

For V9, Samsung has increased its capacity share to about 60% of total V-NAND production, with yield stabilized at over 80%, entering a mature mass production phase. A year ago, V8 still accounted for a significant proportion of V-NAND output, but now its share has dropped below 40%. The speed of production line switching shows Samsung’s quick response capability for AI server demand. In addition, Samsung’s Pyeongtaek fourth factory (P4) dedicated NAND clean room still has more than half of its space uncommissioned, providing ample room for future expansion.

For V10, Samsung officially started mass production in the first half of this year, with less than 5% of total V-NAND output, currently in a ramp-up phase. V10 adopts an approximately 400-layer stacked architecture, with storage density more than 50% higher than V9’s 286 layers, allowing higher capacity SSDs to be fitted in the same CMX modules. Notably, V10 will see the large-scale commercial use of molybdenum material to replace traditional tungsten wiring, reducing wiring thickness by 30% to 40%. Samsung had previously used this in V9 on a small scale.

For V11, Samsung started trial production earlier this year, targeting a stacked layer count of 400 to 500 layers, an increase of about 100 layers over V10. Samsung plans to double the scale of trial production in the second half of the year to accumulate sufficient yield data and accelerate the construction of a mass production system.

Supply concentration effect emerges, consumer market under pressure

The deepening NAND alliance between Samsung and NVIDIA, while boosting Samsung’s performance expectations, is also intensifying supply differentiation across the memory industry.

According to reports, Samsung’s profits in 2026 are expected to exceed the cumulative total from the past 40 years, largely thanks to the structural explosion in storage demand for AI servers. However, on the flip side, when Samsung locks a large proportion of capacity for major tech customers like NVIDIA, NAND supply to other companies and consumers will be clearly squeezed, creating upward risk for storage pricing.

Previously, the AI boom first triggered a DRAM shortage and drove prices up sharply. The NAND market is now facing a similar supply tightening logic. As NVIDIA’s CMX shipment scale continues to grow, this pressure is expected to intensify further.

In the long run, accelerated development of V11 NAND could provide some relief for the consumer market—theoretically, 500-layer stacking technology can enable lower-cost mass production of high-capacity consumer SSDs, but only if Samsung is willing to allocate capacity. Judging by current capacity allocation direction, this outlook remains uncertain.

Lee Jae-yong and Jensen Huang soon to meet, cooperation continues to expand

Cooperation between Samsung and NVIDIA has extended from DRAM and HBM to foundry services, and now NAND flash is also part of the alliance framework, with both depth and breadth of collaboration continuously expanding.

According to reports, Lee Jae-yong will soon meet NVIDIA CEO Jensen Huang in San Francisco. Topics of the meeting are expected to cover supply arrangements for HBM and next-generation NAND products, as well as plans for cooperation on building data centers in locations such as Gwangju, South Jeolla Province, South Korea.

Industry insiders say, "The Samsung-NVIDIA alliance spans storage and foundry businesses, and cooperation discussions between the two leaders will further deepen." With NVIDIA CMX shipments imminent, the strategic significance of this meeting will also become more prominent.

Risk Warning and DisclaimerThe market involves risks; investment needs to be prudent. This article does not constitute personal investment advice, nor does it consider the special investment goals, financial situation or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article suit their specific situation. Invest accordingly at your own discretion and risk. ```