Do More with Less: PowerScale’s Efficiency Advantage in NVIDIA Benchmarks

AI is rewriting the rules of enterprise infrastructure. While the race is on to add more compute, storage has become just as critical—an enabler that keeps data flowing at the speed and scale AI demands. As GPU-powered pipelines expand across thousands of servers, the ability of the storage layer to deliver with efficiency is what determines success. Every rack unit it consumes, every switch it requires, and every watt of energy it draws now shapes not only performance, but also the sustainability and economics of AI at scale.

This is exactly the challenge Dell PowerScale is built to solve. PowerScale F710 is NVIDIA-Certified at the NVIDIA Cloud Partner (NCP) level, delivering the throughput GPUs require while addressing the equally critical dimensions of efficiency—reducing cost, complexity, and energy across the AI data lifecycle.

With the PowerScale F710, organizations gain certified performance backed by measurable infrastructure advantages, helping them scale AI smarter and more sustainably.

What it takes to be a validated NVIDIA Cloud Partner

The NCP Program validates storage platforms that can support the most demanding AI workloads—stress-testing systems for sustained streaming performance, checkpointing, and random I/O across large-scale GPU clusters. Certifications span tiers of up to 16,384 GPUs, simulating real-world AI environments.

Certification proves a platform can keep up with demanding AI benchmarks. The real differentiator, however, is how efficiently that performance is achieved—across hardware, networking, rack space, and power consumption. That’s where architectural design separates scalable infrastructure from expensive overbuilds.

Dell PowerScale advantage

For major service providers, a 16,384-GPU architecture is a standard, large-scale configuration built to support diverse and demanding AI workflows.  Dell PowerScale, VAST Data, and Pure Storage all achieve certified performance at this level*. However, PowerScale delivers this performance with a more efficient configuration.

*Dell PowerScale certifies the PowerScale F710 (Reference Architecture); VAST Data certifies C Nodes and D Boxes in reference architecture available here; Pure Storage certifies its reference architecture with the FlashBlade//S500.

Rack Space Efficiency

PowerScale meets NVIDIA Cloud Partner certification for 16,384 GPUs in just 168 RUs. By comparison, VAST requires ~1.8x more rack space and Pure nearly 5x more to hit the same benchmarks.¹

What this means for customers – lower CapEx and OpEx

• Simplified Configuration
  Where VAST Data and Pure Storage require 28–32 backend switches to support their architecture, PowerScale accomplishes the same workload with just four—an up to 88% reduction¹. That means dramatically less cabling to manage. Customers avoid the sprawl of over-provisioned infrastructure, lowering capital expenses while accelerating deployment and simplifying day-to-day operations.
• Lower Power Consumption
  PowerScale also drives measurable operational savings. Its leaner architecture consumes 41% less power than VAST Data and 72% less than Pure Storage, enabling organizations to scale within power-constrained environments¹. By requiring fewer nodes, switches, and rack units to achieve certified performance, PowerScale lowers ongoing energy costs and advances long-term sustainability goals—without compromising performance.

PowerScale’s efficiency frees valuable datacenter real estate and resources for what matters most—GPUs and compute infrastructure.

The storage engine driving Dell’s AI data platform

While PowerScale’s certification and efficiency is significant on its own, its role extends beyond standalone storage. PowerScale is the storage engine powering the Dell AI Data Platform, recently recognized as a 2025 CRN Tech Innovator, and collaborates with the NVIDIA AI Data Platform shaping the future of AI infrastructure.

The Dell AI Data Platform is built to support every stage of the AI lifecycle—from data ingestion to model training to inference and deployment. At its core, PowerScale provides:

• Parallel File Performance that feeds GPUs at line rate, ensuring clusters stay fully utilized.
• Enterprise-Grade Protection with real-time threat detection, snapshots, replication, and failover to secure critical data pipelines.
• Open Integration with best-of-breed tools including NVIDIA AI Enterprise, Elastic for search and observability, and Starburst for data analytics acceleration.

By combining these strengths, PowerScale ensures that organizations can innovate rapidly, without being locked into proprietary ecosystems or forced into costly architectural trade-offs.

(Read next: The Secret to AI Success? A Smarter Data Platform)

Storage as strategic advantage

As AI workloads continue to expand, storage has become a defining factor in scaling AI—an essential strategic enabler. Meeting these demands requires infrastructure that delivers consistent throughput while minimizing hardware, power, and space.

Dell PowerScale F710 is designed for exactly this balance, it combines certified performance with measurable efficiency gains—requiring fewer components, less power, and a dramatically smaller footprint. In doing so, PowerScale transforms storage from a potential constraint into a catalyst, empowering enterprises to unlock the full potential of their GPU investments.

Connect with a Dell AI expert to learn how PowerScale can help you scale AI environments efficiently and sustainably.

¹ Based on Dell internal analysis of NVIDIA-validated reference designs for 64 SU configurations that adhere to the NVIDIA Cloud Platform Reference Architecture specification for high-performance storage, August 2025.