Why Energy Intelligence Is the New KPI for AI

Every Earth Day, the conversation turns to carbon offsets, renewable energy pledges, and sustainability reports. These should be year-round conversations, and for many people, they are. But this year, a more urgent conversation has popped up—one that’s less about optics and more about operational survival: How much power is your AI actually consuming, and do you have a strategy to manage it?

The answer, for most enterprises, is simply, “No,” or “not a good one,” or, “not a clear one.” That’s a big hurdle, given how quickly AI is developing and expanding. Our recent survey of 300 senior executives at companies with at least $1 billion in annual revenue, produced in partnership with MIT Technology Research, makes the stakes clear: Every single executive surveyed—100%—expects the ability to measure and strategically manage energy consumption to become a core business metric within the next two years. Not a sustainability checkbox. A KPI.

Welcome to the era of energy intelligence.

What is energy intelligence—and why does it matter now?

Energy intelligence is the practice of understanding where, when, and why energy is consumed across your data infrastructure—and using that insight to optimize operations, control costs, and plan for growth. It’s the shift from passive monitoring to efficiency by design.

For decades, energy costs were treated as overhead. They got buried in facilities budgets or lumped into cloud bills. As Everpure Chief Technology and Growth Officer Rob Lee puts it: “Finance picked them up, and nobody really noticed.”

AI changed that. Fast.

GPU-dense AI servers consume an order of magnitude more energy than traditional infrastructure. Training large language models requires extraordinary amounts of electricity. And unlike previous technology waves like mobile and cloud, AI is additive. It doesn’t replace existing workloads. It layers on top of them.

The result: US data centers consumed roughly 4% of national electricity in 2024, a figure that could reach 12% by 2028. And consider this: A single 100-megawatt data center consumes roughly as much electricity as 80,000 American homes. Data centers being built today are gearing up for gigawatt scale—enough to power a mid-sized city.

For enterprise leaders, energy costs have stopped being a line item. They’re becoming a strategic constraint.

The numbers are moving fast, and most organizations aren’t ready

Our survey found that the AI-driven energy cost surge isn’t just coming, it’s already here:

• 68% of executives report their organizations have already faced energy cost increases of 10% or more in the past year due to AI and data workloads.
• 97% expect those costs to keep rising over the next 12 to 18 months—with one in three anticipating growth above 25%.
• Meanwhile, only 22% say their organization feels “very prepared” to handle the mounting costs. The majority—61%—describe themselves as only somewhat prepared.

That gap between awareness and readiness is where the risk lives. And for organizations that hit the wall—when a data center runs out of power or a cloud bill spikes unexpectedly—the disruption to AI initiatives can be severe. As Lee notes: “The average enterprise typically only realizes there’s an issue when they run into the brick wall.”

PUE is a starting point, not a strategy

For years, power usage effectiveness (PUE) has been the go-to metric for data center energy efficiency. While it’s a useful benchmark, it’s increasingly insufficient on its own.

PUE measures how efficiently a data center uses energy: total facility power divided by IT equipment power. A PUE of 1.0 would be perfect efficiency; real-world scores typically range from 1.2 to 1.5 or higher. It tells you how much energy is wasted on overhead like cooling, but it doesn’t tell you how efficiently your actual workloads are running or how your storage, compute, and network choices are shaping your energy footprint.

In AI-driven environments, which most environments now are, that granularity matters enormously. Not all workloads are created equal. Not all infrastructure decisions have the same energy consequences. And as Eric Masanet, professor at UC Santa Barbara’s Bren School of Environmental Science and Management, observes: “Data centers are the backbone of the information economy, and yet when it comes to their impacts, we’re somewhat data-starved.”

Industry coalitions are now working toward standardized metrics that go beyond PUE. The EU’s Energy Efficiency Directive already requires data centers above a certain size to disclose energy performance data annually. True energy intelligence means moving beyond a single number to a full picture of consumption by workload, system, and layer.

Energy’s ‘FinOps’ moment

If the trajectory feels familiar, that’s because it is.

A decade ago, financial operations (FinOps)—the practice of bringing financial accountability to cloud spending—barely existed as a discipline. Enterprises were running up massive cloud bills with little visibility into what was driving them or how to optimize. Today, FinOps is a standard function at most large organizations, complete with dedicated teams, mature tooling, and executive ownership.

Energy intelligence is following the same arc. “Cloud costs illuminated the need for greater financial acumen as a part of technology selection,” says Lee. “I expect the same thing to happen with energy.”

The parallel runs deep. In FinOps, the most consequential decisions happen when you’re choosing your cloud architecture, not when you’re trying to optimize after the fact. The same logic applies to energy. “Once you’ve selected a technology stack, you can optimize around the edges, but if you’ve picked something just inherently inefficient, there’s only so much you can do to clean that up,” says Lee.

That’s why infrastructure selection is where energy intelligence starts, and why storage is often overlooked in the energy conversation.

Storage platforms: The efficiency multiplier

Compute and cooling dominate the energy conversation in AI infrastructure. Storage tends to be an afterthought, but it shouldn’t be. And now we can safely say, not just “shouldn’t,” but “can’t.”

In AI-driven environments, enormous volumes of structured and unstructured data must be stored, accessed, combined, and moved constantly. At that scale, even small inefficiencies compound quickly. And the choice of storage architecture shapes energy consumption in ways that ripple across the entire data center—from direct power draw to cooling requirements to physical footprint.

Three advances in storage technology are making a measurable difference:

• Lower power consumption: The move from spinning hard disk drives (HDDs) to flash-based solid-state drives (SSDs) has dramatically reduced the energy required to store data.
• Improved hardware longevity: Flash-based systems can remain in service two to three times longer than their HDD predecessors—meaning fewer replacement cycles, less logistical overhead, and a smaller long-term footprint.
• Greater power density: Modern flash systems can store roughly 10 times more data in the same physical footprint as legacy alternatives. Fewer devices, fewer storage controllers, fewer enclosures—and significantly less energy required to power and cool them.

Everpure clients have already seen what this looks like in practice: Virgin Media O2 reported a 98% reduction in storage energy consumption after migrating to all-flash infrastructure. British Telecom saw reductions exceeding 90%. THG Ingenuity cut data center power consumption by 80% with no disruption to business operations.

As Lee describes it, the logic is similar to the transition to LED lightbulbs: A single swap doesn’t move the needle, but when enterprises across industries make the shift, the cumulative effect is transformative.

Energy intelligence as a competitive advantage

It’s tempting to frame energy intelligence as a sustainability story. But that framing undersells it.

The organizations building energy intelligence into their operations now are doing something more strategically important: They’re creating optionality. Every kilowatt saved on storage and cooling is capital that can be redirected into the next wave of AI innovation. Organizations that compute more while consuming less will have more room to scale, lower operating costs, and a structural advantage over competitors still running inefficient legacy infrastructure.

There’s also a resilience dimension. As grid constraints tighten and cloud costs rise, companies with deep visibility into their energy footprint will be better equipped to keep their AI initiatives running—and better positioned to avoid the disruptions that will increasingly hit organizations that waited.

Reputation matters too. “It still matters to people that they’re consuming products and services from an organization that takes sustainability seriously,” says Everpure Field CTO for EMEA Patrick Smith. “A reputational component is probably where energy efficiency stands up best as a competitive advantage.”

The survey data makes the imperative clear:

• 74% of leaders are already working to optimize existing infrastructure
• 69% are partnering with energy-efficient cloud and storage providers
• More than half are implementing AI workload scheduling and investing in more efficient hardware

But for organizations still early in this journey, the starting point is measurement. You can’t optimize what you can’t see. Building the capability to track energy consumption by workload, by system, and across your full infrastructure stack—including third-party cloud and managed services, where 71% of executives say rising costs originate—is the foundation everything else depends on.

This Earth Day, the most meaningful thing an enterprise can do for the planet is to build the visibility to actually know where its energy is going and possess the infrastructure discipline to do something about it. The AI race is already underway. The organizations that win it won’t just be the ones who compute the most. They’ll be the ones who compute the most efficiently. Energy intelligence isn’t a sustainability initiative. It’s how you stay in the game.