News & Insights

The Grid Has Become the Bottleneck

Across North America, the electric grid is entering a period of sustained scarcity. AI and cloud workloads are driving electricity demand to unprecedented levels. Data centers already consume an estimated 4.4% of U.S. electricity, a share projected to reach up to 12% by 2028, with AI servers alone accounting for much of the incremental load. Meanwhile, interconnection queues exceed 2.3 terawatts, more than double the size of the entire U.S. generation fleet. In many regions, new data centers are waiting five or more years for a grid connection, and utilities are imposing deferrals, moratoriums, and phased load caps just to preserve reliability.For the first time in the digital era, access to power, not land, capital, or demand for compute, has become the primary constraint on growth. The consequence is a new kind of scarcity: one defined not by cost but by availability of megawatts. Efficiency can no longer be measured solely by how little energy we waste, but by how productively we use what little we have.

Why PUE is no Longer Sufficient

For nearly two decades, Power Usage Effectiveness (PUE) has defined how the data center industry measures efficiency. It was simple, universal, and transformative, driving an entire generation of operators to reduce overhead through better cooling, distribution, and design discipline.

This strategic focus yielded results: the industry-wide average PUE dropped from approximately 2.5 to the 1.5 range. At the leading edge, hyperscalers like Google, Amazon, and Microsoft now report averages of 1.1 or lower, often achieved through evaporative cooling. However, PUE is a metric designed for an era of abundance. While it effectively measures infrastructure overhead - powering IT versus cooling - it fails to capture how efficiently the full envelope of available energy is converted into useful computation.

Today, that’s the central question: how do we maximize the productivity of every megawatt?

Introducing PUx: The Next Generation of Metrics

To address this challenge, Verrus has introduced the PUx family of metrics, extending PUE to capture total resource productivity:

• Provisioned Utility Capacity (PUC): the percentage of total energy from the utility interconnect that can be allocated for SLA-backed IT nameplate capacity (analogous to 1 / design-day peak PUE). In other words, how large is the “container” available for fully guaranteed compute workloads relative to the utility feed?
• Productive Utilization of Nameplate (PUN): the percentage of the nameplate power reserved for IT equipment that is actually consumed, i.e., how fully the allocated capacity container is utilized by productive IT load.

Together, PUE, PUC, and PUN form a more complete lens on efficiency, utilization, and productivity in an era when every megawatt matters.

Case Study: How Verrus Gets More From Every Watt

Metrics are only as powerful as the designs they measure. The Beyond PUE paper details how Verrus’ PowerFlow™ architecture translates this framework into measurable gains. A modeled 70-MW Verrus data center designed with PUx in mind achieved:

• PUC = 77 % ➜ utility watts allocatable for IT equipment
• PUN = 96 % ➜ nearly full utilization of that IT allocation
• ➜ ~75 % of total available utility energy converted into productive IT load

By comparison, a modeled conventional design converts only 49 %, meaning Verrus’ architecture delivers a 53 % improvement in overall energy productivity, equivalent to unlocking an additional 18 MW of compute within the same utility interconnection.

That translates to an effective value of ~ $35MM per year based on an average lease rate of $145/kW per month at an assumed 96% PUN.

What Makes It Possible

Verrus’ architecture reimagines how energy is distributed across the entire facility. Rather than reserving large chunks of unused capacity for rare peak conditions, PowerFlow™ dynamically orchestrates the energy flow so that every watt contributes to productive work. Put simply, it expands the effective “container” for compute relative to the utility feed and enables higher utilization of that container without sacrificing reliability or uptime. Verrus achieves these gains through an integrated suite of technologies and design philosophies:

Better Capacity Utilization with PowerFlow^TM

This is the intelligence behind Verrus' design. PowerFlow™ combines a novel electrical distribution architecture with advanced optimization, controls, and telemetry to unlock far greater utilization of leased IT nameplate capacity.

More Capacity with CapacityBoost^TM

Verrus integrates dispatchable energy resources, including cooling optimization, thermal storage, and Battery Energy Storage Systems (BESS), to harness daily load variations, trim peak consumption, and shift grid power demand, unlocking additional standard SLA-backed capacity from a given utility feed.

Supercharged Capacity Utilization with HyperBoost^TM

Beyond standard SLA-backed capacity, Verrus can optionally deliver reliable performance by coordinating day-ahead workload shedding only during peak cooling system strain. For customers that choose to participate, this gives access to otherwise unavailable capacity 95-99% of the year.

Together, these innovations show how PowerFlow™ fundamentally changes the capacity equation. Instead of simply reducing overhead, Verrus unlocks more usable IT headroom, extracts more value from the same interconnect, and delivers higher utilization nearly every day of the year. The table above summarizes the step-by-step gains compared to a conventional design.

The Bottom Line

The analysis points to an important shift in how efficiency must be measured and achieved in digital infrastructure. Traditional metrics like PUE capture how efficiently energy is converted, but not how completely capacity is utilized. As grid constraints tighten and AI workloads surge, that distinction becomes critical. The PUx framework provides a more complete lens, one that reflects both how energy is used and how fully interconnect capacity is translated into productive compute.

Viewed through this lens, the opportunity for improvement lies not only in better cooling or power conversion, but in maximizing total system utilization. Designs optimized around this principle deliver more deployable compute per contracted megawatt—expanding usable capacity without additional interconnections.

At Verrus, this principle is embedded from the ground up. PowerFlow™ architecture exemplifies what next-generation efficiency looks like: higher productive utilization, faster time-to-capacity, and reduced grid impact. More broadly, PUx offers a framework for an industry that scales sustainably:  doing more with every watt in an age of power scarcity.

Sign up below to receive the technical report.

Ready to see how much more your megawatts could deliver?

Click here to contact Verrus about leasing capacity that gets the most out of your dollar.