The New Load Frontier: Insights from E3’s Kush Patel

Teri Viswanath

February 4, 2026

Key points

• Communities driven by transparent regulatory processes and strong utility planning have successfully managed the unprecedented scale and speed of large load growth — without unfairly burdening existing customers.
• Historical growth models are out. Innovative forecasting, planning, and infrastructure investment plus flexible collaboration and lifecycle risk management are essential.
• The next phase of data center growth will be shaped less by incentives and more by power, water and community acceptance.

In the contemporary energy landscape, few developments have stirred as much discussion and concern as the expansion of data centers and other significant electricity-consuming entities. With the increasing necessity for robust digital infrastructure and cloud computing, the United States faces complex questions about managing supply, costs and strategy. Data centers — central to our modern interconnected world — present unique challenges for infrastructure investment, rate design and long-term planning.

To unravel these vital issues, CoBank’s energy economist Teri Viswanath sat down with Kush Patel, a senior partner at E3, and asked him to share his insights from the work his firm has been engaged with in helping regulators, utilities and data centers find equitable paths for energy access. This conversation distills the most pressing questions facing these actors, offering a lucid, in-depth exploration of the topic.

Teri Viswanath: Kush, about a year ago, your team tackled the issue of whether data centers are pushing up costs or shifting them to other consumers, notably in Virginia. E3 was engaged by the Joint Legislative Audit & Review Commission in Virginia to examine the impacts of data center growth on the state’s electric infrastructure needs and associated costs, as well as the distribution of these costs across customer classes. Can you walk us through what you discovered from that research and its implications?

Kush Patel: Certainly, Teri. The issue of cost impacts from data centers is at the forefront of public discourse, and it’s easy to see why. Data centers represent immense, concentrated electricity demand, and naturally, communities worry about how this will affect their bills. Our analysis in Virginia aimed to clarify whether these concerns were substantiated.

What we found is that, while data centers do exert pressure on local grids, a well-designed regulatory framework and proactive cost management can go a long way in mitigating negative outcomes. Effective rate design — where the costs associated with serving these large loads are allocated appropriately — can prevent unfair cost shifts to smaller consumers. Furthermore, we observed that data center growth often prompts utilities to invest in infrastructure, which, if managed wisely, can actually benefit the entire grid. In Virginia, by rigorously examining local impacts and policy responses, we demonstrated that the narrative of runaway bills isn't inevitable. Transparency and collaboration between all stakeholders are critical to ensuring that the benefits and burdens of data center growth are shared equitably.

Viswanath: When “large loads” are discussed in the utility space, the definitions seem to vary. Yet, the return of significant load growth in the U.S. is mostly driven by these so-called large loads — primarily by commercial (data centers) and industrial sectors. In fact, the nation is now realizing growth rates not seen since the 1990s, with the expansion particularly pronounced in Texas, PJM and Georgia. But, as I mentioned, the definition of “large load” varies widely, with data centers ranging from micro (150 kW) to hyperscale (100+ MW), and utilities experiencing requests from 25 MW to over 250 MW.

How should stakeholders understand what constitutes a large load, and why does it matter?

Patel: That’s a key point, and one that is often overlooked. “Large load” is not a standardized term — it’s entirely contextual. In one region, a large load might be a data center with 50 megawatts of demand; elsewhere, a manufacturing facility drawing only 10 megawatts may qualify. Defining what makes a load “large” depends on local grid conditions, regulatory thresholds, and economic considerations.

Why does this matter? Because the way we categorize these loads affects everything from how utilities plan for growth, to how costs are distributed, to how policy is crafted. In areas experiencing rapid data center development, it becomes essential to forecast future demand under multiple scenarios — what happens if growth plateaus, accelerates, or shifts geographically? These forecasts underpin infrastructure investment and help determine rate structures that are both fair and sustainable.

Moreover, recognizing the diversity among large loads — not just data centers, but other commercial and industrial actors — ensures that solutions address real needs, not just hypothetical ones. In short, precision in defining “large load” unlocks clarity for all stakeholders, leading to better decisions.

Another point I would like to make is that 3% growth (the level of growth that is predicted to occur today) in the '60s, '70s or even the ‘90s was on a much smaller base. So, although we’ve witnessed similar growth rates on a percentage basis, the absolute growth we are talking about is unprecedented. It is also occurring in a very compressed period of time. Consider the most amount of supply our nation has ever interconnected in a single year and then double that number for the next decade…this is the scale of planning and development that is now required at the utility level, at that state policy level, and the regional operator level, etc. in order to hit the targets we are talking about.

Viswanath: It feels like there is also an added layer of complexity in forecasting large load demand, especially with speculative inquiries and multiple site explorations by the same customer, which can lead to double-counting. Correlated risks exist as large tech companies often make coordinated decisions affecting multiple sites simultaneously. Utilities must differentiate between real and inflated demand to effectively manage resources and investments, right?

Patel: I think it is clear to everyone that load growth is accelerating. Utilities are projecting record load growth and investing heavily in new generation and transmission. Yet, what remains unclear is just how long this trend will continue, especially when we consider that efficiency is probably not included in these projections, right?

We believe that forecasting needs to shift from a single point-in-time prediction to ongoing measurement, testing, and adjustment that incorporates operational and financial feedback.

I should also mention that there are important externalities that will influence demand growth. I have previously mentioned that power supply will prove to be a limiting factor in just how many data centers will ultimately get built. The industry will ultimately catch up but this is not the only factor to consider.

For illustrative purposes, we developed a matrix to identify how key considerations for siting a data center have evolved and will evolve over time. Tax incentives and access to fiber drove the initial wave of development, and power availability is now squarely in the driver’s seat. But unfolding now and soon to play a more decisive role is community opposition and water availability. These factors will likely carry greater weight in where future data centers get built than access to sufficient power supplies.

Viswanath: The polling for our webinar showed that about a third of the audience has had direct experience with data centers or other large loads, while others have not. Why is this conversation relevant to everyone in the energy sector?

Patel: The relevance is universal, even for those who haven’t yet encountered large loads directly. Data centers are emblematic of broader shifts in energy consumption patterns, and their impact extends across the entire system. As demand clusters and intensifies in certain regions, everyone — utilities, regulators, and other commercial entities — must adapt.

For those not yet approached by large loads, understanding the dynamics at play prepares them for future developments. Rapid technological change means that what seems remote today could become immediate tomorrow. The lessons learned in places like Virginia serve as case studies for other regions, offering templates for effective regulation and stakeholder collaboration.

Energy professionals must stay informed and engaged, because the strategies developed for large loads will shape grid reliability, affordability, and sustainability for years to come.

Viswanath: There is real public anxiety around data centers driving up electricity bills. Having worked with data centers and the utilities that serve them, what is your nuanced view on this subject, and what should communities keep in mind?

Patel: The anxiety is palpable, and the headlines can be alarming. Yet, the real story is more layered. Yes, data centers demand a lot of energy, and yes, this can put pressure on local infrastructure. But thoughtful rate design, clear cost allocation and policy oversight are powerful tools to keep bills in check.

Our research consistently finds that communities with transparent regulatory processes and strong utility planning are able to absorb these new loads without unfairly burdening existing customers. In fact, data centers can spur beneficial investment in grid modernization, resilience, and renewable energy integration.

Communities should demand transparency, ask hard questions about cost allocation, and insist on policies that reflect local realities. When everyone acts as a stakeholder — utilities, data centers, consumers, and regulators — solutions emerge that balance economic growth with consumer protection.

Viswanath: I would like to get into a few particulars on how electric cooperatives can manage the financial risks associated with serving large loads. Your team published a report analyzing credit policies as viewed from the data center or customer’s perspective. Can you talk about this?

Patel: Sure, we published a white paper a few months ago that was sponsored by the Data Center Coalition, looking at best practices with credit and collateral requirements. The report considers infrastructure risk from both a data center (or customer perspective) as well as through a utility lens.

Our team observed that financial risk is often viewed on what I would call a la carte perspective — meaning that risks are evaluated separately rather than considered holistically and from a lifecycle perspective.

For a customer who's coming in with a large load request and doing all the pre-permit, pre-construction, construction, post-COD. What does that look like from their perspective? Then, of course, what does it look like from a utility perspective? It's not symmetrical. The utility may not be taking on a whole lot of risk initially until those projects become more real as they advance through the interconnection and signing process.

I think there are opportunities or defined points along a project’s timeline when mutual de-risking can take place. Where utilities can be more proactive in thinking through and understanding where that customer's coming from and potentially allow for, I don't want to say leniency, but just different flexibilities or different options to lessen the risk exposure.

For example, if a data center customer has a co-location model and needs to go out and get a tenant, the utility might allow the flexibility for that to happen first, rather than imposing relatively strong collateral requirements that would force that potential customer to go somewhere else. Ultimately, that result could prove unfavorable for both parties. But, if the utility was collaborating on a viable project with a credit-worthy customer that simply needed another 30 days to get to the next milestone or finish line, perhaps that flexibility would pay off. I think those are some of the things that we see as being mutually helpful, with transparency and communication absolutely critical for evaluating risks.

Again, it is clear that everything is evolving in this space in real-time. From my perspective, it's always helpful to be transparent and understand the risk and what both sides are trying to solve for to get to that win-win situation where no party is taking on too much risk or shifting risk inappropriately.

Viswanath: There was an interesting point you raise about understanding the life cycle of risk, can you expand upon this?

Patel: I mentioned the idea of maintaining project discipline without overburdening early stages. I would also add that jointly developing a milestone-based structure offers a clear roadmap for when and why financial requirements apply to the project.

Credit requirements often combine multiple tools (letters of credit, contribution in aid of construction, guarantees, deposits, etc.) to cover various risks. While each serves a purpose, applying them without a clear framework can create unnecessary redundancy. This overlap can lead to overcollateralization, tying up customer capital, slowing or canceling projects and increasing attrition risk.

Viswanath: I noticed that your group recently published A Guidebook of Industry Best Practices and Examples from Real-World Amazon Data Center Case Studies to better understand whether existing rate designs adequately protect ratepayers from cost increases. Can you provide us with a few important takeaways from that work?

Patel: That work presented results from evaluating several Amazon data centers across a diverse set of utilities, representing different geographies and market structures. One of the more reassuring findings is that the Amazon data centers we evaluated generate revenues that meet or exceed their cost to serve. Coming full circle to where we started this conversation is that large loads, especially high load factor customers like data centers, can provide benefits to the system beyond their revenue contribution.

With regard to this recent analysis, we developed a set of overarching best practices for utilities and regulators to consider when designing rates and tariffs for large loads. There are a couple important takeaways that are worth sharing:

• Charging data centers at least marginal cost will help ensure no customer cross subsidization by other rate payers.
• To generate surplus utility revenue, data centers can be charged a portion of nonmarginal, or fixed, costs as well, which could lead to a range of equitable outcomes, including putting downward pressure on rates.
• Given uncertainty in data center load growth, tariffs should balance risk management mechanisms with other priorities (e.g., economic development and shared benefits), the perceived risk from the data centers’ perspective (e.g., utility failing to interconnect on schedule), and avoid being punitive or discriminatory.

Viswanath: Looking ahead, what do you see as the next big story or challenge in the intersection of data centers, large loads, and the energy market?

Patel: The future holds several converging challenges. Grid reliability will remain front and center as data centers proliferate, especially in regions where infrastructure is aging or insufficient. I expect to see innovative approaches to energy sourcing, from on-site renewables to grid-scale battery storage, as both utilities and data centers strive for resilience and sustainability.

There’s also a growing emphasis on demand response — using smart technologies to modulate consumption in real time — and on forging public-private partnerships to address emerging threats and opportunities. Over time, the conversation will move beyond costs to broader questions of climate responsibility, energy equity, and the evolving role of digital infrastructure in our society.

The key is adaptability. Energy markets are changing quickly, and those who remain open to collaboration and innovation will shape the next headlines, not just react to them.

For more information on E3, visit https://www.ethree.com

To learn more about CoBank’s Guide for Serving Large Loads, visit https://www.cobank.com/knowledge-exchange/power-energy-and-water/a-guide-for-serving-large-loads