AI, Energy and the Future of Efficient Data Center Operations

Rising AI workloads are expected to drive higher data center energy use, costs and emissions. Some experts even predict data center energy consumption could double by 2030, placing added strain on already burdened power grids. As a result, reliable, resilient and affordable energy has become a top priority for data center operations worldwide.

To address this challenge, we've been asking ourselves - what if AI could help improve power use, enhance grid flexibility and streamline industrial processes, turning today's challenges into solutions?

Rethinking Data Center Operations to Align with Power Demand and Supply

At Dell Technologies, it starts with improving efficiency across our AI portfolio - from next-gen servers that lower CPU power consumption by up to 65%,¹to pioneering cooling innovations and advancing software management systems to improve optimization. Customers like Verne, Fulgent Genetics and Emirates NBD have been able to rethink their infrastructure and power strategies, significantly reducing their footprint.

We're also looking for where we can push boundaries and go beyond infrastructure innovation to uncover smarter ways to align power demand with supply at the workload level.

One of our latest explorations, Concept Astro, uses agentic AI, digital twins and automation to spur smarter decision making to improve IT operations and energy consumption. Consider an interconnected, adaptive AI Factory that can forecast the time it'll take to complete a task and predict the energy requirements to do so. Imagine it's grid-aware, pulling in real-time grid data and forecasts to identify the ideal moment, location and resources needed to run the job, prioritizing workloads based on business needs, energy availability and/or cost. Users could choose from a menu of options and recommendations or workloads could be set to run automatically within pre-configured parameters.

[Link]

Now add in the ability to access all this information through a simple dashboard or by requesting reports customized to your role - from researcher or data center operator, to CFO, CIO or facilities. For those running workloads, Concept Astro could help you make smarter decisions as it pulls in real-time insights for a comprehensive view across - and beyond - your data center(s). For facilities and the CFO, it's a way to see how your teams are lowering costs by running energy intensive workloads at optimum times and locations without compromising business outcomes.

[Link]

This capability is exciting but right now just a concept, with our engineers pushing the boundaries of what is possible to deliver at scale. In this iteration, we're using digital twins to simulate workloads so we can extract insights and identify ways to use available data center power more effectively with power grid awareness. Concept Astro builds on the Dell AIOps monitoring suite and features the capabilities offered today through Dell's AIOps Assistant - including customers being able to ask context-aware questions about their connected infrastructure and receive tailored recommendations based on their current system configuration.

Real-World Impact Through Public-private Collaboration

To test how far Concept Astro can go, we've been working with Scripps Institution of Oceanography at the University of California San Diego to expand their laboratory capabilities for coral reef research. Using the Dell AI Factory with NVIDIA, the team is converting millions of underwater images into high resolution 3D models to assess the health of coral reefs across the world. With around 350 GB of images collected per dive and the team completing anywhere between 300-400 dives a year, processing these images requires significant compute and storage.

During a recent pilot, Concept Astro helped Scripps identify the best times and location to run workloads based on cost, speed, emissions or a combination of all. The team was able to schedule workloads during optimum energy windows, reducing strain on the grid without disrupting research continuity. This enabled Scripps to achieve cost savings of 20% and lower emissions by 32%.² In addition, by upgrading their legacy data center equipment to the Dell AI Factory with NVIDIA, including Dell's latest generation of servers, Scripps can now process twice as many images in the same amount of time.

Building a More Resilient Energy Ecosystem

At Dell, we're helping customers unlock the full potential of enterprise AI while minimizing energy consumption and operational costs. We're pushing ourselves to continually innovate, refine and enhance our offerings to deliver even greater business and societal value.

What drives us is helping our customers overcome their biggest challenges. With energy-efficient technologies already transforming operations and forward-thinking concepts like AI-powered workload scheduling, we can help enterprises to reduce costs, enhance grid resilience, and drive meaningful progress.

We'll continue to share updates as we explore making data centers more grid aware and efficient. In the meantime, check out Chief AI Officer John Roese and Chief Strategist Vivek Mohindra's latest discussion on AI, power and cooling - and watch this space for exciting updates from Dell Technologies World, May 19 - 22, 2025.

¹ Based on Dell analysis comparing the SPECint and SPECFP scores of the AMD EPYC 5th Gen 9965 in a Dell R7725 (2980 and 2350) with the same scores for an Intel Xeon 8280 in a Dell PowerEdge R740XD (375 and 296).

² Based on Dell Technologies Concept Astro pilot with Scripps Institue of Oceanography at the University of California San Diego. Data collected December 2024 - March 2025.