[datacenter_tag_image]
On March 5, the House Committee on Energy and Commerce Subcommittee on Energy held a hearing to address the challenges of meeting rising power demand. With unprecedented load growth upon us, I’d like to share my perspective on navigating this critical period.
Increasing demand
Electrification, reshoring of manufacturing, and the explosive rate of data center expansions due to AI are primary drivers of load growth. Though efficiency gains are expected in these areas, the sheer magnitude of the forecasted increase means efficiency improvements are insufficient. The Jevons Paradox—where efficiency gains drive increased consumption—may even undermine these very gains.
Within 15 years, PJM Interconnection’s summer peak demand is projected to surge to 220 GW, up from its all-time high of 165 GW in 2006. This growth is equivalent to adding nearly four New York Cities annually. This trend of surging peak demand is a nationwide challenge.
Limited supply
Matching demand may be tricky given unit retirements, generator interconnection backlogs, and the various challenges for new supply to come online, such as potentially triggering the need for grid expansions. These grid expansions or upgrades can be time-consuming and can make the interconnecting generation no longer pencil on costs.
Flexibility minimizes frictions to better align supply and demand
In February 2025, Duke University published its “Rethinking Load Growth” report, and Tyler Norris, a report author, testified at the hearing.
The report finds that by allowing for minimal curtailment or onsite generation, the US could support the increased power needs of AI data centers. Specifically, approximately 98 GWs of new load could be integrated into the grid with a minimal average annual load curtailment rate of just 0.5%. This means that for a tiny fraction of the year, new loads would need to be reduced, but for the vast majority of the time, they could run at full capacity.
In the sections below, I’ll share some clips from Tyler Norris’ testimony along with my analysis.
“Nearly half of our power infrastructure sits unused…”
The grid is predominantly planned to meet firm peak conditions, which is very limited duration, less than 5% of the year.
In that vein, I do not think loads necessarily need to be studied and planned for firm 24/7 continuous power, which may trigger costly and time-consuming transmission expansions. By recognizing the up-front value of a load’s ability to drop from the grid in the transmission planning process and before load interconnection, loads can take “non-firm” interruptible service and allow grid operators to control their onsite dispatchable generation and/or batteries to manage the relatively infrequent annual peaks (<5% of the year).
The result is faster load interconnection, higher utilization of the existing grid, and “right-sizing” transmission.
Recognizing the value of firm commitments to be flexible in the transmission planning process allows loads to avoid triggering costly, lengthy grid upgrades that are only needed for a relatively small number of peak hours annually, enabling faster interconnection.
“Flexible load strategies refer to… for a very very small number of hours a year… consumers could reduce their consumption… like if they have their own on-site power…”
Flexible load strategies help to provide a bridge to address the “today” issues of getting connected quickly while grid planners are hard at work on the “tomorrow” issues like planning for the construction and integration of new transmission and new centralized supply.
Dispatchable onsite power, such as firm gas generators or batteries, give grid planners the controllable certainty needed to implement these flexible load strategies. Grid operators and planners can rely on real, tangible and power producing assets co-located with load instead of taking years to build more infrastructure to serve the very small number of hours a year when these loads may pose a reliability risk. This dispatchability also enhances the integration of intermittent renewable resources like wind and solar.
The way it is done today is analogous to ordering a checked luggage bag to replace your carry-on and waiting years for it to arrive, just to fit one extra pair of socks.
While more grid infrastructure will ultimately be needed in the future, we should not ignore the solutions technologically and economically available today to get loads connected quicker and to mitigate impact to ratepayers.
“If new loads utilize the existing system more, that can result in actual reduction in rates…”
Keep investment risk with the asset owners and not on captive ratepayers who may be left holding the bag (stranded costs) if demand does not materialize. We can align the interests of large loads and the grid by incentivizing loads to bring their own peak power and contribute to grid reliability without necessitating grid upgrades.
The more we use it, the cheaper it gets.
“If load or generator is willing to be flexible… it can connect to the grid much more quickly…”
“The biggest opportunity may be from these large centralized new loads… especially if they already have their own on-site power and they’re able to run that for a few hours a year, then we could add a lot more of that load a lot more quickly.”
“you’re going to have on-site natural gas power plants already, so, let’s not have redundant capacity so that we’re (not) going to have to install new gas power plant(s) and all the transmission to deliver the power and in response that data center could get online even more quickly…”
I think these last three clips are good takeaways. By integrating the value of flexibility commitments into regulatory and planning frameworks, we can avoid triggering the lengthy and costly process of building long lead time infrastructure already plagued by supply chain bottlenecks, permitting, and ratepayer cost concerns.
We must advocate for a load’s firm commitment to be flexible as an allowable planning solution. Rather than build new near-term capacity (both transmission and centralized generation), we can optimize meeting demand with local, firm, onsite dispatchable generation, often already readily available in the form of customers’ reliability focused onsite generation.
Leveraging customers’ dispatchable onsite generation for grid services and as alternatives/complements to new required transmission can accelerate load interconnection, reduce ratepayer costs, and optimize utilization of the grid we have today.
There are lots of complexities involved with the Supply/Demand balance – reducing the frictions between them is key. I think the best way to do that right now is to recognize the full potential of loads’ dispatchable onsite generation.
This article was originally published on LinkedIn.
