Supermicro Launches Petascale All-Flash Storage Server with NVIDIA Grace CPU for Advanced AI Applications

Supermicro Unveils Advanced AI Storage Solution

At the forefront of technological innovation, Supermicro, Inc. has recently introduced a groundbreaking all-flash storage server designed specifically for high-performance software-defined storage workloads. As revealed during the GTC 2025 Conference, this system leverages the NVIDIA Grace CPU Superchip, marking an exciting new chapter for IT solutions tailored for artificial intelligence (AI) and machine learning (ML).

Supermicro's latest storage server is poised to transform the landscape of high-density storage solutions. With its architecture catering to advanced AI training, analytics, and enterprise storage demands, this innovation exemplifies Supermicro's commitment to delivering cutting-edge technology that enhances performance without compromising efficiency. The dynamic collaboration with NVIDIA and WEKA has resulted in a product that enables businesses to optimize their AI factories for superior operational power.

Harnessing the Power of NVIDIA

The new Petascale storage server is built around the NVIDIA Grace CPU, which boasts an impressive 144 Arm Neoverse V2 cores, delivering exceptional I/O performance for demanding software-defined storage requirements. According to Charles Liang, the company's president and CEO, this collaboration emphasizes Supermicro's dedication to offering innovative solutions and superior customer choices.

Utilizing Supermicro’s Building Block Solutions® strategy, the newly introduced ARS-121L-NE316R storage system features a symmetrical architecture that is perfectly suited for scaling AI workloads. The powerful combination of Supermicro’s engineering expertise and NVIDIA’s technology facilitates the system's scalability while fully harnessing its PCIe Gen5 performance capabilities. With support for 16 hot-swap EDSFF PCIe Gen5 E3.S NVMe drives, the system can achieve a staggering 983TB raw capacity. This means an entire rack containing 40 systems can provide up to 39.3PB of raw storage—a monumental achievement for businesses that require vast and efficient data management solutions.

Energy Efficiency at Its Core

One of the standout features of Supermicro's Petascale storage server is its energy efficiency. By working closely with NVIDIA, Supermicro ensures that customers can reap the benefits of high-performance storage with reduced power consumption compared to traditional x86 server systems. Rob Davis, vice president of storage networking technology at NVIDIA, emphasized this important aspect, noting how innovative hardware choices align with various application requirements.

Furthermore, Supermicro's partnership with WEKA has crystalized into a formidable offering for AI, HPC, and enterprise data management workloads. The well-optimized combination of the 144 Arm cores along with an integrated 960GB of LPDDR5X memory on the CPU platform allows for maximum performance when pairing with the WEKA® Data Platform, which boasts a zero-copy architecture tailored for high efficiency.

Real-World Applications

Initial trials conducted on the Supermicro Petascale storage server with the NVIDIA Grace CPU have yielded remarkable results. Nilesh Patel, Chief Product Officer at WEKA, highlighted that the performance scaling and power efficiency exhibited during the testing were significant advantages for enterprises focused on AI and ML applications. The collaboration between these technology giants underscores a shared vision: to empower businesses to efficiently manage and analyze colossal data sets in a sustainable manner.

Presentation at GTC 2025

For those keen to learn more about this innovative storage solution, Supermicro and NVIDIA plan to showcase the Petascale storage server and its capabilities at the NVIDIA GTC 2025 Conference, deepening insights into how it can revolutionize business operations concerning data storage and AI applications.

In conclusion, Supermicro's introduction of the petascale all-flash storage server marks a pivotal moment for organizations leveraging AI technologies. With high-performance, efficiency, and scalable architecture fueled by NVIDIA's technological prowess, businesses can look forward to tackling tomorrow's challenges head-on. For further details, visit Supermicro’s dedicated webpage on their storage solutions for AI.