Just as electricity, water, and roads defined the infrastructure build-out of the last century, digital infrastructure has become the critical foundation of this century, powering the everyday conveniences of modern life. As demand for data centers grows to sustain this essential utility, the question is no longer simply how quickly we can build. It is how we build infrastructure that works for the utility grid and the communities around it.
That is why Verrus is helping launch the AI Energy Management Alliance (AEMA), alongside founding members Emerald AI, Google, and NVIDIA. We’re joining a broad list of partners, from AI and semiconductor companies to utilities and power producers, all working toward the same goal: building data centers that don’t just connect to the grid but work with it.
For Verrus, that starts with a simple belief: data centers can be good citizens.
Constraints Have Changed, And So Must The Grid
The primary constraint on data center deployment has shifted from capital and chips to grid power. As AI accelerates demand, developers are colliding with an outdated interconnection process. Traditional grid planning models still evaluate requests under worst-case assumptions, requiring capacity reserves to be held 365 days a year as if the facility were drawing its absolute maximum load on the hottest, highest-stress day on the grid.
A flexible data center can operate differently. A facility can adjust its power draw in response to grid conditions by shifting computing activity, drawing on storage, using paired generation, or a combination of these. Instead of making a data center a fixed load the grid has to build around, flexibility allows it to become a resource the grid can rely on when demand is highest.
That is the opportunity Verrus is helping AEMA pursue. The alliance is bringing together the AI and power industries to advance flexible data centers, develop practical interconnection approaches, and establish standards that recognize measurable grid value. This type of flexibility should be the standard, and it’s what we’ve been building for since day one.
Flexibility Only Counts If The Grid Can Rely On It
Utilities and grid operators need to know what a facility will do before it connects, when it will respond, how quickly it can respond, and how that performance can be verified. AEMA is working to answer those questions through common performance metrics, technical requirements, operating data, and faster interconnection pathways for facilities that make credible commitments.
Verrus has had this in mind from the beginning. We pair a power flow management system with long-duration battery storage to reduce import from the grid when the system needs it. Our 70 MW platform was rigorously tested at the National Lab of the Rockies, formerly the National Renewable Energy Laboratory, a U.S. Department of Energy national laboratory, where it demonstrated the ability to curtail its grid draw to as little as zero in under a minute following a utility signal, and transition to battery-powered operation without disrupting IT workloads.
On the hottest day of the year, when demand across the grid is at its highest, we can be the load that steps back.
Making Better Use Of The Grid We Already Have
AEMA demonstrates that through joint support, we don’t need to build new infrastructure to meet growing demand. More than 100 gigawatts of capacity sits idle on the U.S. grid outside of peak periods, and much of that infrastructure was built to handle a small number of the highest-demand hours each year.
Flexible data centers can help make better use of that existing capacity. Every gigawatt of capacity unlocked through flexibility is capacity that doesn’t necessarily require a new power plant or transmission line to serve it. That can mean faster access to power for new data centers while making more efficient use of infrastructure that ratepayers have already paid for, delivering more compute without asking more from the grid.
Being A Good Grid Citizen Means Being A Good Neighbor
Responsible data center development extends beyond connection to the power system. From water management and noise mitigation to traffic and infrastructure, every facility must respect its environment. That’s why we design our data centers from the ground up to protect the grid, preserve vital resources, and actively support the communities we call home.
Our closed-loop cooling system cuts water use by 99% compared with legacy evaporative cooling data centers, bringing annual water use per site to less than that of a few sit-down restaurants. We use batteries for primary power backup instead of diesel generators. We conduct an independent noise study for every site to ensure our facilities are designed to operate well below local limits, day and night. We also put hard limits on power, water, and noise into writing before we break ground and publish our performance numbers over time.
Because Verrus owns and operates its data centers, we’re invested for the long term in the communities where we build and in making sure our projects remain good neighbors for decades to come.
Building The Standard For What Comes Next
AEMA provides companies, utilities, and power leaders a place to work through the practical questions standing between flexible data centers and grid interconnection: what should a flexible data center be required to deliver? How should that performance be measured? How can utilities evaluate flexibility during interconnection? And how can the system make better use of existing capacity while maintaining reliability?
The answer cannot simply be to add more demand and build more infrastructure around it. We need data centers designed to thoughtfully work with the systems they depend on. That means flexible power use, careful water use, and designing for the community that will be home to a data center for decades.
Verrus is bringing what we’ve already built and tested to that effort. As a launch partner in AEMA, we’re working alongside leaders across the AI and energy industries to turn flexible data centers into something utilities can plan around, interconnection processes can recognize, and the industry can deploy at scale.