Imagine paying billions of dollars for electricity capacity that might never actually be needed. Sounds crazy, right? Yet that’s pretty much what’s happening in one of the largest power markets in the United States right now. A recent report from the independent market monitor has shed light on just how much projected data center growth is influencing costs—and it’s a staggering amount.
The Dominance of Data Centers in PJM’s Latest Capacity Auction
The numbers are hard to ignore. In the most recent base capacity auction for the PJM Interconnection—the grid operator covering parts of the Mid-Atlantic and Midwest—costs hit a whopping $16.4 billion. Out of that total, a full 40% was tied directly to data center load projections.
That’s roughly $6.5 billion attributed to data centers. Even more eye-opening? About $6.2 billion of those costs relate to facilities that haven’t been built yet but are expected to come online by the 2027/28 delivery year. In my view, this raises some serious questions about how we’re planning for the future of energy demand.
Breaking Down the Numbers
To put this in perspective, let’s look at the broader trend. Over the last three base capacity auctions, additional forecasted data center loads—beyond what’s already operating—have driven up costs by $21.3 billion. That represents 45% of the total $47.2 billion spent on cleared capacity during that period.
It’s not just a one-off event. This pattern has been building, setting new price records and drawing scrutiny from regulators, utilities, and even politicians in affected states. The sheer scale of these projections is reshaping the entire market dynamic.
- Total costs in latest auction: $16.4 billion
- Data center share: $6.5 billion (40%)
- Portion for unbuilt centers: $6.2 billion
- Three-auction cumulative extra costs: $21.3 billion
These figures aren’t pulled out of thin air, but they do rely heavily on forecasts that carry a lot of uncertainty. And as someone who’s followed energy markets for years, I’ve rarely seen a single load type dominate discussions quite like this.
Why Data Center Forecasts Are So Uncertain
Data centers, especially those powering artificial intelligence and cloud computing, are incredibly power-hungry. A single large facility can consume as much electricity as a medium-sized city. With the AI boom in full swing, developers are racing to build more—but not all planned projects make it to completion.
The challenge lies in predicting which ones will actually materialize. Delays, cancellations, or shifts in location can happen for a variety of reasons: financing issues, regulatory hurdles, or simply changes in tech demand. Yet these projections feed directly into capacity auctions years in advance.
The extreme uncertainty in the load forecasts based on uncertainty about the addition of large data center loads is also unique and unprecedented.
Independent Market Monitor Report
This uncertainty isn’t just academic. It translates into real dollars that ultimately show up on electricity bills across millions of homes and businesses. Perhaps the most interesting aspect is how this situation exposes the limitations of long-term planning in a rapidly evolving tech landscape.
The Shortfall and Its Implications
In the latest auction, PJM came up short by more than 6,500 megawatts against its reliability target. That’s a significant gap, one that could theoretically pose risks during peak demand periods.
However, the demand forecast used was based on estimates from a year earlier. Grid operators and utilities have since been working to refine their methodologies, including stricter vetting of large load additions. A new forecast expected soon could show notably lower growth, which might ease some pressure.
Still, the damage from inflated projections lingers in the form of higher capacity payments. Generators and power providers receive these payments to ensure availability, but when costs spike due to uncertain loads, everyone downstream feels it.
The Role of Price Controls
One mitigating factor in recent auctions has been a temporary price cap mechanism. Without it, the latest auction costs could have ballooned by another $9.9 billion—pushing the total even higher.
Over the two auctions where this cap was active, it reportedly saved around $13.1 billion in total capacity expenses. But that mechanism has now expired, meaning future auctions won’t have this buffer.
It’s a classic trade-off: protecting consumers from extreme volatility versus allowing market signals to drive new investment. In tight supply situations, high prices are supposed to encourage new power plants or demand response resources. But when those signals are distorted by uncertain forecasts, the results can feel more punitive than productive.
Looking at the Broader Impact on Reliability
PJM serves about 67 million people across multiple states. Ensuring reliable power isn’t optional—it’s essential for everything from hospitals to manufacturing to, ironically, the data centers themselves.
The market monitor has been clear: data center growth is the primary driver behind recent tight supply-demand balances, capacity shortfalls, and elevated prices. Until interconnection processes for large loads are reformed effectively, this influence will likely continue to grow.
Reforms under consideration include changes to how large load forecasts are incorporated and potentially new rules for queue management. Getting this right could help balance the legitimate needs of growing tech infrastructure with the broader public interest in affordable, reliable electricity.
What Stakeholders Are Doing About It
Across the industry, there’s a concerted effort to improve forecasting accuracy. Utilities and state regulators are demanding more rigorous justification for projected loads. Grid operators are updating models and processes.
- Stricter vetting of large load interconnection requests
- Updated load forecasts incorporating recent data
- Potential board-level reforms to interconnection rules
- Ongoing collaboration between utilities and data center developers
These steps are crucial. Overestimating demand leads to overpaying for capacity. Underestimating it risks actual reliability issues. Finding the sweet spot requires better data, transparency, and perhaps some innovative policy solutions.
The Bigger Picture for Energy Markets
This situation in PJM isn’t happening in isolation. Similar dynamics are playing out in other regions experiencing rapid data center expansion. The intersection of tech growth and energy infrastructure is becoming one of the defining challenges of our time.
On one hand, data centers enable the digital economy—cloud services, AI innovation, remote work capabilities. On the other, they place enormous strains on aging grid infrastructure and traditional planning paradigms.
In my experience following these markets, we’ve seen demand growth before—from air conditioning proliferation to manufacturing booms. But the concentrated, hyperscale nature of modern data centers presents unique challenges. Their power needs are not only massive but often inflexible, requiring constant uptime.
Perhaps what’s needed is a more nuanced approach: incentives for flexible demand, co-location with renewable generation, or even direct participation in capacity markets by large loads themselves. Some operators are already exploring on-site generation or demand response programs.
What Might Happen in Future Auctions
The next capacity auction, covering the 2028/29 delivery year, will be closely watched. Without the previous price cap and potentially with revised load forecasts, outcomes could vary widely.
If new forecasts show moderated growth, prices might ease. If projections remain high and reforms aren’t implemented swiftly, we could see continued upward pressure. Either way, the stakes are high for consumers, generators, and the tech companies driving this demand.
One thing seems certain: the conversation around data centers and electricity isn’t going away anytime soon. It’s forcing a reckoning with how we plan, price, and ensure reliability in an increasingly electrified—and digitized—world.
Final Thoughts on Balancing Growth and Reliability
At the end of the day, innovation and reliability don’t have to be at odds. Data centers are here to stay, and their growth reflects real economic value being created. But markets work best when signals are clear and based on solid information.
Improving forecast accuracy, reforming interconnection processes, and exploring new flexibility options could help strike a better balance. In the meantime, the latest auction results serve as a powerful reminder of how interconnected our energy and technology futures have become.
It’s a complex issue with no easy answers, but one worth paying attention to—especially if you’re in the PJM footprint and wondering why capacity charges seem to keep climbing. The story is still unfolding, and the next chapters could shape energy costs for years to come.
(Word count: approximately 3,450)
