Yes Energy recently released its updated long-term outlook for PJM in GridSite Nodal, its fundamental-based node-level modeling platform. The forecast reveals that PJM’s long-term market conditions are changing rapidly across multiple fronts.
In recent years, PJM's long-term load outlook has shifted sharply, with the regional transmission organization (RTO) now forecasting an unprecedented increase in electricity demand, driven largely by data centers, other large loads, and electrification. Meanwhile, retirements and new generation additions are reshaping the grid’s resource mix and the marginal resources that set prices. However, the impact of these changes varies by location as load growth, retirements, and new resources occur unevenly across the region.
Transmission upgrades will also have a growing impact on the system over the long term, alleviating constraints in some areas while redirecting others. At the same time, greenhouse gas prices and renewable portfolio standards (RPS) will reshape system costs and buildout, as well as net-zero commitments.
This article digs into the trends set to shape PJM through 2046, the forecasting methodology behind Yes Energy’s long-term nodal price forecast, and what these shifts mean for developers, utilities, energy buyers, and investors navigating the market.
Between 2027 and 2046, PJM’s total installed capacity by type will evolve significantly according to Yes Energy’s latest long-term forecast. Together with changes in load, transmission, and policy, that evolving mix shapes three key outcomes — prices, congestion, and emissions.
The equilibrium scenario, which is Yes Energy’s base case, predicts coal plant retirements will increase by 2030 and be replaced with new, more efficient thermal generation. A significant amount of solar and some wind will also come online in the near term, based on analysis of the region’s interconnection queue.
Looking out to 2046, the model shows the 2030 shift playing out on a larger scale, with even more combined cycle generation coming online to meet load growth. The overall mix is still largely thermal, but solar, storage, and wind will grow significantly over the 20-year horizon.
These shifts will have a significant impact on prices, congestion, and emissions. PJM’s average real-time locational marginal price (LMP) in real 2024 dollars is expected to rise 82% to $99 per megawatt-hour in 2046.
One of the primary factors driving LMPs is a 92% increase in aggregate real-time congestion costs, which rise to $6.4 billion by 2046. Total PJM CO2 emissions will rise 64% to 566 million tons, while PJM’s average locational marginal emission rate (LMER) falls by 5%. LMER measures the change in total power system carbon or other pollution emissions resulting from a small increase or decrease in demand at a specific location and time.
PJM’s annual load will rise about 72% by 2046, predominantly driven by data centers. The econometric load, which you can think of as the base load, accounts for the largest share of the total load in 2027. Electric vehicles (EVs) and data centers account for a relatively small percentage of the overall load. As we move forward in time, the econometric and EV loads grow, but more gradually.
PJM data center load growth, however, explodes in the near term, more than doubling from 2027 to 2030, and more than doubling again between 2030 and 2046.
Location is the key to understanding why LMPs diverge so sharply across PJM. For example, Dominion and ComEd sit on opposite sides of the system, with two critical interfaces — AP South and Bedington–Black Oak — between them.
In a perfect world with no transmission constraints, every pricing point across PJM would be the same. But the long-term forecast shows congestion accounts for 95% of the price spread between Dominion and ComEd, driven almost entirely by those two interfaces. Line losses contribute just $2 per MWh.
As load grows over time, the system gets tighter, driving LMPs up significantly over the next 20 years. Dominion, which is the primary electric utility in Northern Virginia, powers the country’s largest data center market. Its real-time LMP will reach $116 per MWh in 2046, exceeding PJM’s average by $17 per MWh. ComEd’s real-time LMPs, on the other hand, will be $13 per MWh lower than the RTO’s average.
The largest contributor to congestion in PJM is AP South, with $38 billion in cumulative costs across the forecast horizon. This is because the transmission system can’t keep pace as we try to move power into Virginia to meet the data center load.
As noted earlier, emissions in PJM are projected to grow about 64% by the end of the forecast horizon, but the LMER falls by roughly 5%. The reason for this is that locational marginal emissions rates follow the marginal resource responding to the next increment of load, not the average generation fleet. This makes LMER a great way to consider the impact of new resources on total system emissions.
You can see how this plays out by comparing ComEd’s and Dominion’s generation mix with PJM as a whole. ComEd is still a very nuclear-heavy region in 2046, while renewables and combined cycle plants dominate Dominion’s generation mix. System-wide, PJM’s generation mix sits in the middle, with a meaningful nuclear component and a large coal/combined-cycle/peaker block. Renewables represent about a quarter of the mix.
To build market-specific equilibrium scenarios and inform assumptions around future market changes, the Yes Energy modeling team updates its assumptions around load forecast, generation fleet, fuel prices, availability, policy, interchange, transmission topology, and more every quarter.
These assumptions inform Yes Energy’s nodal capacity expansion and nodal production cost models, which produce detailed forecasts of locational marginal prices, locational marginal emission rates, revenues, congestion, and power flows at specific locations.
Unlike hub-level forecasts, which aggregate pricing across broad regions and can obscure location-specific dynamics, or forecasts built on machine learning and regression techniques that are inherently tied to historical patterns, Yes Energy's approach runs fully nodal simulations of the power system. This preserves the fidelity of the underlying transmission network and produces forecasts that reflect the actual physical and economic drivers of price formation at each node, rather than statistical relationships extrapolated from the past.
This approach also makes Yes Energy forecasts defensible. A defensible long-term forecast shouldn't just explain what; it should explain why. Yes Energy’s market forecasts are built from detailed, transparent assumptions rather than simply extrapolating historical data.
For market participants, this distinction has real consequences: nodal-level detail and grounding in fundamentals translates into more defensible inputs for valuation, procurement, siting, and risk management decisions.
In other words, Yes Energy doesn’t just give you a forecast; it shows you the assumptions and the physical market conditions that create the forecast, so you can better understand the risks and make better investment decisions.
In addition to the equilibrium scenario, the Yes Energy modeling team runs a low-regulation scenario, which is more thermally dominant, and an infrastructure expansion scenario, which assumes higher renewables penetration. Comparing these comprehensive scenarios reveals how sensitive an outcome is to different assumptions.
The value of a nodal price forecast lies in how it informs decisions. By combining forward-looking price forecasts with detailed market and historical data, GridSite Nodal enables you to identify the locations that matter, understand what drives prices and congestion, and assess how these factors may affect a project's economics.
The analysis can then be tested against historical plant performance and different market scenarios, giving you a clearer view of potential revenues, profitability, and risk. In this way, the forecast becomes more than a projection of future prices. It’s a tool for evaluating where to invest and how resilient those decisions may be under different market conditions.
To learn more about GridSite Nodal and see a real-world example of how you can use the tool to conduct a nodal analysis, watch the on-demand webinar.