Performance increased by 30 times! Nvidia (NVDA.US) and SK Hynix develop AI solid-state drives to be mass-produced in 2027

The Zhitong Finance App learned that according to reports, Nvidia (NVDA.US) and South Korea's SK Hynix are collaborating to develop a solid-state drive (SSD) with artificial intelligence as the core, using NAND chips. This project is codenamed “Storage Next” within Nvidia, and is called “AI-NP” (AI NAND Performance) within SK Hynix, and aims to break the bottleneck on the storage side of the current AI computing power system.

This jointly developed AI SSD can be called a leapfrog upgrade in terms of performance indicators. Its core goal is to achieve up to 100 million input/output operations per second (IOPS), which is about 10 times faster than today's top enterprise SSDs, and can even improve performance by up to 30 times in specific environments.

The function of this device is to act as a key “middle layer” between high-bandwidth memory (HBM) and traditional high-capacity storage. Through extremely high throughput and extremely low latency, it ensures that massive data in the AI inference process can be instantly recalled by the GPU, thereby significantly improving generative AI response speed and reducing overall energy consumption.

In terms of technical support, SK Hynix and its partner SanDisk (SNDK.US) are jointly promoting a new standard called HBF (High Bandwidth Flash). This vertical stacking structure modeled after HBM will provide the necessary hardware foundation for this AI SSD.

According to the schedule disclosed by both parties, the first product prototype is expected to be launched in the second half of 2026, and official mass production is scheduled for 2027. At that time, this new storage solution is expected to be deeply integrated with cutting-edge hardware such as Nvidia's future Rubin CPX series GPUs.

This partnership not only marks that Nvidia is further strengthening its AI ecosystem's vertical integration capabilities, but also indicates that SK Hynix is trying to establish a new hegemony in the AI NAND storage field after leading the HBM memory market.

However, this trend has also raised concerns in the supply chain. The industry predicts that with the explosion of AI inference demand, the NAND flash memory market may repeat the supply shortages and price fluctuations experienced by DRAM.