PJM Just Cited the Compute Heat Rate in Their Landmark Market Reform Proposal.

The largest wholesale electricity market in North America just proposed redesigning itself, and included the CHR demand-side pricing framework. Here's what that means.

I’m floored by this.

This week, on May 6^th, 2026, PJM published a 70-page white paper called “Powering Reliability Through Market Design.” It was signed by their new CEO, David Mills. It is one of the more significant market reform proposals in PJM’s history – and it specifically mentions and proposes the Compute Heat Rate¹ as part of the market solution(!!).

The paper asks three questions:

Should reliability be rationed?

Should the capacity market be replaced by an energy market?

And: what happens when the fastest-growing load class on the grid has a price tolerance 127 times higher than anything the market was designed for?

That last question is the CHR and potentially changes everything.

The problem PJM is trying to solve

For two decades, PJM’s capacity market worked. Slow load growth, predictable retirements, gas plants that could be built in three years. The market’s three-year forward auction was calibrated to a world where supply and demand moved at roughly the same speed.

That world is gone.

Data centers now want to interconnect faster than generation can be built to serve them. The capacity auction has cleared at its administrative cap three times running. The 2027/2028 auction fell 6,500 MW short of the reliability target. Construction costs for a new combined cycle plant have doubled to $2,100-2,300/kW. And the interconnection queue is growing faster than the study process can clear it.

PJM’s diagnosis: this isn’t a temporary supply crunch. It’s a structural mismatch that will persist for a decade or more.

The credibility trap

Here’s the part that matters for anyone with an energy contract, a generation asset, or a rate case pending.

In the paper, PJM calls it the “credibility trap.” High capacity prices signal scarcity. But because most load is unhedged against those prices, the rate shock triggers political intervention, and regulators cap prices. Investors thus see the cap and discount future revenues, which means capital stays on the sidelines and the shortage persists.

The price signal is correct. But the system can’t sustain it long enough to trigger the investment the signal is supposed to attract.

Sound familiar? The same structural problem Peter Perri identified in ERCOT forward curves last week is playing out in PJM’s capacity market. The mechanism is different (capacity auctions vs. energy forwards), but the root cause is identical: the market has no framework for pricing a demand class that doesn’t respond to the signals the market was designed to send.

Three paths, one missing variable

PJM proposes three reform paths:

Path A: make the capacity market financially durable through mandatory long-term hedging. Force every load-serving entity to come to the auction pre-hedged so that scarcity prices can clear without causing political crises.

Path B: ration reliability explicitly. If the grid can’t serve everyone at the historical standard, decide who gets curtailed first. New large loads that connected without bringing supply go to the front of the curtailment queue.

Path C: shift revenue recovery from the capacity market to the energy market. Raise scarcity price caps, let energy prices reflect the true value of reliability, and require long-term energy contracts to protect consumers from volatility.

Each path has tradeoffs. PJM doesn’t recommend one. But all three paths share a common dependency: they need a demand-side pricing framework that tells you which loads will respond to scarcity prices and which won’t.

That’s where the paper pivots.

The Hyperscale Paradigm Shift

On page 49, the paper introduces a section called “The Hyperscale Paradigm Shift.” This is the analytical climax of the entire document, and it’s all about the Compute Heat Rate.

PJM observes that data center load is “fundamentally different from the residential and commercial load that defined the grid of the 20th century.” It identifies three levers of inherent flexibility: workload shifting, on-site generation and storage, and advanced load control architectures that can curtail demand by 10-30% in seconds.

Then it asks the key question: at what price does that flexibility activate?

At the current PJM energy price cap (~$3,700/MWh), the answer is: it doesn’t. The economics of curtailing compute operations are “generally unattractive for most AI workloads.”

At $10,000/MWh, the answer changes. And this is where PJM introduces the Compute Heat Rate.

CHR as market infrastructure

PJM’s language is precise:

They introduce the Compute Heat Rate as “the maximum electricity price at which a given AI workload remains profitable, a demand-side analog to the gas heat rate familiar from resource dispatch modeling.”

The blended CHR across workload types is approximately $6,300/MWh: “roughly 127 times the current wholesale average but squarely within the range that could be accessed under achievable scarcity pricing reforms.”

Then PJM maps the CHR distribution against their proposed scarcity pricing levels:

At $10,000/MWh, commodity AI tasks and enterprise contracted jobs (CHR $800-1,270/MWh) become economic to curtail. Mid-tier inference workloads (CHR ~$8,000/MWh) approach their threshold. Frontier AI operations (CHR exceeding $50,000/MWh) continue uninterrupted.

PJM calls this “precisely the graduated, price-elastic demand response the grid needs: not a binary shutdown, but a tiered response where the most flexible compute responds first, in proportion to price.”

Their proposed reserve market reform (the Ramp/Uncertainty Reserve ORDC at $1,000-1,900/MWh) “begins to create the signal necessary to activate this flexibility.” And: “a higher cap ceiling would reach further into the CHR distribution.”

Why this matters

The gas heat rate tells you the floor: the minimum price at which a gas plant breaks even.

CHR is the ceiling. The maximum price at which the fastest-growing load class on the grid will continue to consume electricity without curtailing.

PJM just said they need both.

For generators: your assets in high-penetration hubs are worth more than the curves say, because the demand-side ceiling is orders of magnitude above the supply-side floor.

For C&I consumers: the load class that’s driving your capacity costs up will never voluntarily curtail at any price the grid has historically produced.

For regulators: moratoriums and price caps don’t solve the structural problem; scarcity pricing calibrated to the CHR distribution does.

For investors: the spread between the gas heat rate and the CHR is the tradeable opportunity. Hubs where data center penetration crosses the threshold will reprice. Hubs that don’t, won’t.

The paradox

The paper closes the CHR section with a sentence that deserves to be read twice:

“Paradoxically, this could mean that the influx of the very loads contributing to the resource adequacy problem today will help to shrink the relative scale of the missing money problem over time.”

The load everyone is blaming for breaking the grid is also the first load class in history with a quantifiable, graduated, price-elastic response at scarcity levels. Traditional residential and commercial load has a vertical demand curve: perfectly inelastic, no price response, no curtailment signal. Data center load, through CHR, has a sloped demand curve at high prices. That slope is what makes market-based scarcity management possible.

And now the largest wholesale electricity market in North America is proposing to redesign itself with the CHR front and center. Awesome!

Hans Royal is the originator of the Compute Heat Rate™ (CHR) framework. All views are his own and do not represent those of any employer or affiliated organization.

1. Royal, Hans, The Compute Heat Rate: Quantifying AI-Driven Electricity Price Tolerance
and Its Implications for Wholesale Market Repricing (February 28, 2026).
Available at SSRN: http://dx.doi.org/10.2139/ssrn.6322318