AI servers housing a boatload of NVIDIA GPUs were previously able to handle tons of temporary context data known as Key-Value (KV) cache, but the astronomical scale of this data has started bottlenecking the chips, necessitating the need for Context Memory Storage (CMX), which is exactly where Samsung comes into play.
A new report states that the Korean giant is already supplying tons of V9 NAND to NVIDIA for this purpose, with V10 undergoing mass production. While these shipments should remove the context data limitations from GPUs, it’s other companies and consumers that’ll bear the brunt, as there will be little to no supply left.
Samsung is also researching V11 NAND for NVIDIA’s CMX, capable of reaching an impressive 500 layers of vertically-stacked storage
To sate demand, Sedaily reports that Samsung’s V-NAND production covers a capacity of 100,000 units for NVIDIA, with 60 percent of the company’s capacity allocated to the V9 line. The V10 NAND should allow for higher storage layers, ensuring that NVIDIA can outfit higher-capacity SSDs within the same CMX module.
At present, each Context Memory Storage for AI servers can house 576 SSDs, reaching a whopping capacity of 9,600TB. While this can potentially eliminate the data bottleneck for NVIDIA GPUs, it comes at the cost of a supply deficit, which pretty much guarantees that other companies and consumers are drowned in misery as they’ll look to fulfill their measly storage requirements at inflated prices.
Initially, the AI boom triggered a DRAM shortage, causing prices to skyrocket, and this cataclysmic change has already adversely impacted the storage market. To remind you, this is just the start. As AI GPUs begin offloading more and more temporary context data to CMX, NVIDIA’s ravenous demand will rise exponentially.
According to the report, storage chips for CMX will increase from 35 million terabytes to over 100 million terabytes in 2026, with Samsung expected to reach 250 million terabytes in NAND shipments, cementing its position as NVIDIA’s potential chief supplier. On the plus side, after years of battling losses, the Korean technology giant has finally found its calling, with its 2026 profit said to exceed cumulative earnings generation in the past 40 years.
With this industry shift, NAND manufacturers have already realized where the lucrative partnerships lie, but that’s where the problems begin to exacerbate for the entire industry. The only silver lining we can see from Samsung’s and NVIDIA’s alliance is the acceleration of V11 NAND, which features a maximum stacking height of 500 layers, making it 100 layers more than V10, which already offered 50 percent higher storage density than V9’s 286 layers.
This technology can be leveraged to mass produce consumer-grade SSDs with larger capacities at a lower cost, but only if Samsung chooses to allocate capacity. From the looks of it, catering to the rest of the industry appears to be a fever dream.
News Source: Sedaily
About the author: Omar Sohail is a reporter and analyst for Wccftech's mobile section, specializing in the technology and business of the mobile industry. His expertise lies in the intricate hardware supply chain, covering developments in semiconductor manufacturing, chip lithography, and camera sensor technology.
Follow Wccftech on Google to get more of our news coverage in your feeds.
