ANALYSIS: PJM, Artificial Intelligence, and the Bill We Pass On to Our Neighbors

6.3 billion behind the counter

The number on the meter

An electricity auction may seem as cold as a table of numbers. The one PJM unveiled on July 14, 2026, is not. It organizes the capacity that will need to be available from June 2028 to May 2029 in thirteen U.S. states and the District of Columbia. It is therefore setting the stage today for tomorrow’s bill.

The independent auditor Monitoring Analytics attributes approximately $6.3 billion of the $16.4 billion associated with this auction to demand from data centers. This is an estimate, not an accounting verdict already itemized line by line on every bill. But the sheer magnitude of the figure now makes it impossible to comfortably deny the reality.

Progress with Someone Else’s Address

Artificial intelligence loves to talk about the future. Electricity, on the other hand, requires a billing address. When a new, massive demand enters a shared market without a perfectly isolated mechanism, the cost doesn’t stop at the fenced-off perimeter of the digital campus. It spreads.

It reaches suppliers, distributors, regulators, and then households and businesses, according to pricing rules that vary from one region to another. The exact share paid by each party remains unknown. The direction of this transfer, however, has become a politically explosive issue.

The cloud has never been light. It simply weighed on someone else’s power outlet.

What PJM Actually Bought

A promise of availability

The capacity market doesn’t just sell electrons consumed in the moment. It compensates resources so they will be available when the grid needs them most. PJM secured 138,318 megawatts of non-firm capacity, in addition to 10,864 megawatts acquired by regions with firm resource requirements.

This technical language masks a collective insurance policy. We pay to reduce the risk that a demand spike will meet with insufficient supply. Yet this insurance is now being put to the test by growth that is anything but ordinary: new data centers may emerge with concentrated, continuous, and immense power demand.

A Price at the Cap

The price reached the authorized cap of $325 per megawatt-day. It is 2.5% lower than the cap from the previous auction, set at $333.44, but the apparent decline does not tell the whole story. The total remains at $16.4 billion, as in the previous cycle.

The price cap has kept prices in check. It hasn’t created the missing power plants, shortened the waitlists for grid connections, or slowed demand. We’ve held the thermometer; we haven’t cured the fever. That’s why celebrating a 2.5% drop would be nothing more than a paper victory.

A price cap can provide protection. It can also reveal just how hard the market is already pushing against the ceiling.

Scarcity Amid Abundance

6,831 megawatts below the standard

Despite the billions committed, PJM reports that the total capacity secured remains 6,831 megawatts below its reliability requirement. The resulting installed margin is 14.4%, which is 5.6 percentage points below the 20% target. This is the second consecutive cycle in which the auction has ended well below this margin.

PJM clarifies that such a shortfall does not mean it will be unable to meet demand in 2028–2029. Rather, it signifies thinner reserves and an increased risk compared to the standard of an expected load loss one day in a decade. The distinction is crucial. It is not reassuring.

The American Paradox

The grid has just committed $16.4 billion and remains below its reliability standard. This is the stark paradox of this auction: paying a huge amount without buying enough. Consumers are therefore not financing abundance. They are financing administered scarcity.

This scarcity existed before the rush toward AI. Power plants are closing, projects are waiting to be connected, and transmission lines take years to approve and build. But the rapid arrival of large loads is adding weight to a framework that was already creaking.

Prices are rising because the future is arriving faster than the wires capable of carrying it.

Why Data Centers Are Driving Up Prices

A load unlike that of a home

A household turns lights on and off, sleeps, cooks, and uses heating or air conditioning. A large data center requires a massive and stable power supply, often 24 hours a day. When a cluster of similar projects concentrates in the same region, the demand curve doesn’t just rise gradually. It changes shape.

The PJM auction report notes that the regional reliability requirement has increased by 3,613 megawatts, including a 1,374.5-megawatt rise in the load forecast largely attributed to new large loads. That’s not the whole problem. It’s a documented surge.

Industrial Time vs. Digital Time

A server can be ordered, installed, and replaced much faster than a power plant, transmission line, or substation can navigate the process of studies, permits, appeals, financing, and construction. The real conflict, therefore, is not between technology and electricity. It is between two different timeframes.

Digital capital promises months. The electrical infrastructure responds in years. In between, the capacity market transforms the physical delay into a price signal. Then that signal seeks out a portfolio. Political power begins exactly at this point.

We can speed up a calculation. We cannot compress a high-voltage line with the push of a button.

The Mechanism of Transfer

A collective bill, unequal contracts

PJM itself warns that the 16.4 billion does not correspond to the total cost directly paid by all consumers. Certain costs are covered by self-generation or bilateral contracts and are not exposed to auction prices in the same way. This is precisely why we must resist overly simplistic slogans.

According to Monitoring Analytics, the figure of 6.3 billion reflects the impact of data center demand on capacity costs. It does not specify which households will receive how much, in which state, or on what date. This distribution will depend on local regulations, supplier purchases, and decisions by public utility commissions.

The Unwitting Neighbor

But uncertainty about the distribution does not eliminate the moral issue. If a new industrial project drives up the price of a shared insurance policy, everyone who buys in that market may feel the ripple effect. The neighbor becomes a financial partner without having signed off on the project.

This is the political crux of the matter. Elected officials can tout the investment, the construction jobs, and the strategic power of AI. They must also explain why the energy risk posed by these projects should be pooled with that of a grocery store, a school, or a residential building.

Socializing a grid is necessary. Socializing an industrial boon is never automatically necessary.

The Giants’ Promise and the Market Test

Protecting the consumer

In March, Google, Meta, Microsoft, and other data center companies promised the White House they would protect consumers from price hikes linked to their energy and infrastructure needs. The commitment sounds simple. The market described by Monitoring Analytics makes it much less so.

When high demand enters regional forecasting and then influences the quantity and price purchased for everyone, a private promise is not enough. Contracts, guarantees, a calculation method, and recourse are needed if costs are nevertheless passed on to other customers.

The difference between paying for your energy and paying for your connection

A data center can purchase renewable energy while still being responsible for other costs: capacity, transmission, interconnection, local grid reinforcement, and reserves. So saying “we pay for our electricity” does not answer the whole question. We must ask who pays for the grid connection.

Monitoring Analytics proposes removing data centers from the common capacity market and supplying them through a dedicated auction. The idea may be debatable in its details, but it’s sound in its intent: to make visible the cost that the current architecture obscures.

A promise isn’t protection as long as the bill can still find a way around it.

The Temptation of the Single Culprit

The grid was already aging

Blaming data centers alone would be convenient but wrong. U.S. electricity prices are also affected by fuel costs, transportation and distribution investments, the replacement of aging equipment, extreme weather events, and varying state policies. Fortune helpfully points out this complexity.

The new major loads are therefore not the root cause of every price increase. However, they act as a catalyst in regions where supply is growing more slowly than demand. Political maturity lies in holding these two ideas together without using one to overshadow the other.

Responsibility Does Not Require Exclusivity

A player does not need to be the sole cause to have to pay its share. This is, in fact, the age-old trick of all opaque systems: everyone points to other factors, and then no one assumes the marginal cost they add. The consumer, however, cannot break down their bill into excuses.

The most honest assessment, therefore, remains twofold. Yes, the network has been burdened by decades of sluggishness and underinvestment. Yes, Monitoring Analytics estimates that the growth of data centers has added 6.3 billion to this specific bill.

Complexity explains the bill. It must never be used as an excuse to shirk responsibility.

The Backstop and the Race for Capacity

Buying After the Auction

Faced with the shortfall, PJM is working on a last-resort procurement mechanism. The goal is to secure new resources capable of bridging part of the gap and maintaining reliability. Such an intervention acknowledges that the regular auction did not yield the desired quantity.

The mechanism may attract projects or secure longer-term contracts. It may also add costs. Everything will depend on its structure, the level of actual competition, commissioning guarantees, and how the beneficiaries of the new demand will contribute to the payment.

Capacity cannot be mandated

A back-up auction does not magically produce a turbine in a field. It does not open a transmission corridor or issue a permit. It purchases an additional commitment in a world where many energy commitments are already behind schedule.

The Federal Energy Regulatory Commission has also approved an expedited process capable of reviewing up to ten applications per year for interconnection of large new or upgraded resources, set to expire at the end of 2027. This is an attempt to speed up the development timeline, not a guarantee of results.

The backup plan becomes dangerous when it offers reassurance before construction has even begun.

The choice governors can no longer avoid

Welcome them, but under what conditions?

States want the campuses, the investments, and a place in the AI value chain. Rejecting all new digital infrastructure would be a step backward. Accepting it without cost conditions would be a policy of submission. Between the two lies the difficult task: negotiating before the triumphant announcement.

Governors can demand commitments to in-house production, dedicated capacity contracts, financial guarantees, ramp-up schedules, and protections for residential customers. They can also coordinate their demands to prevent individual states from competing by footing the bill for others.

Investment Blackmail

Some will argue that overly strict rules will scare projects away. That’s possible. But a project whose viability depends on the silent transfer of its energy risk deserves to be scrutinized. Investment is not a sacrament that absolves itself of responsibility.

The PJM grid crosses state lines. A decision made in Virginia can alter planning and prices far beyond its borders. Federalism then becomes an invitation to shared responsibility, not a license to privatize the benefits and regionalize the costs.

Competition between states must not become an auction of the assets of those who are absent.

Low-Income Households Facing Industrial Policy

The Small Increase That Isn’t So Small

A higher electricity bill doesn’t mean the same thing to everyone. For a well-off household, it might be an annoyance. For a family already struggling, it competes with groceries, rent, medication, or transportation. You don’t need to imagine a specific person to understand this math.

Prudence dictates that we not assign a specific amount to these households just yet. Final rates will vary. But waiting for the final notice of a rate hike before addressing equity would be tantamount to accepting that all major decisions have already been made in advance.

The Hidden Regression

The kilowatt-hour doesn’t check income before costing more. A uniform rate hike therefore takes a heavier toll on the budgets of those with the least. When industrial policy is channeled through the bill for an essential service, it risks becoming a regressive tax without actually being called one.

Rate commissions will need to break down costs, test demand forecasts, review contracts, and protect vulnerable customers. Fairness won’t be decided solely by grand speeches about AI. It will be decided in regulatory documents that few citizens will read.

The most political bill is often the one presented as purely technical.

The Cost of a Flawed Forecast

If the campuses come

If the projected demand materializes, PJM will need additional capacity, transmission lines, and flexibility. Failing to prepare for this would expose the region to higher prices and reduced reliability. Underestimating the load would be irresponsible.

But preparing the grid still does not answer the question of how to share the burden. An accurate forecast can lead to an unfair policy if costs are allocated without distinguishing the primary beneficiaries of growth.

If the campuses don’t materialize

The other risk is less dramatic: building for projected demand that is delayed, reduced, or shifted. Infrastructure committed based on uncertain projects can become stranded costs. Here again, contracts determine who is left holding the bag.

We must therefore require deposits, verifiable milestones, and obligations proportional to the reserved capacity. The network cannot treat every project announcement as guaranteed consumption, nor every withdrawal as an accident to be absorbed collectively.

Predicting is not the same as believing. It is about managing the cost of error.

A national strategy without an energy balance

AI as Power Infrastructure

The United States is right to view computing capacity as a strategic issue. Advanced models, cloud services, research, and military or industrial applications all depend on considerable computing power. Abandoning this ground would carry a real geopolitical cost.

But a serious national strategy must take into account data centers, production, transmission, water, equipment, and timelines. Otherwise, Washington is celebrating a digital capacity that regional networks must improvise with tools designed for a bygone era.

Power That Won’t Admit Its Price

Technological sovereignty cannot be funded by transfers that no one dares to name. If AI is a national priority, its energy infrastructure must also become a national priority, with transparent funding, defined responsibilities, and public safeguards.

The opposite is a clandestine industrial policy: the government encourages, companies build, PJM plans, commissions allocate, and then households discover the result in their monthly bill.

A grand strategy that hides its costs is still just a trendy slogan.

What Regulators Must Make Transparent

Causality, not just the total

Upcoming decisions should publish scenarios with and without major loads, show the reinforcements triggered by each group of projects, and distinguish between energy, capacity, transmission, and distribution. Without this level of detail, anyone can pick the figure that fits their narrative.

Monitoring Analytics has announced a more detailed analysis of the auction. This will be essential, as the figure of 6.3 billion currently comes from calculations provided by its president to the media. A robust independent estimate deserves to be examined, replicated, and challenged on a transparent basis.

The Contract of Truth

Making information visible does not mean passing judgment in advance. It means enabling the data center, the supplier, the regulator, and the citizen to discuss the same cost. Transparency is not a moral bonus. It is the minimum infrastructure required for a legitimate bill.

We must also monitor the implementation of supply projects. A winning capacity that arrives late does not protect the grid as intended. A load that increases without an equivalent contribution exacerbates the imbalance. Both sides of the auction must be held to their promises.

What is not separated in the accounts always ends up mingling in anger.

Canada isn’t standing on the sidelines

The Same Rush, Other Grids

Quebec, Ontario, and other provinces are also courting digital investments while managing limited electricity supply, connection waitlists, and competing industrial needs. The pricing model differs from PJM’s, but the dilemma crosses the border: who gets the power, at what price, and at the expense of what other use?

Historically affordable electricity is becoming a strategic advantage. It’s also becoming a scarce political resource as the electrification of transportation, heating, and industry collides with demand from data centers.

The Lesson Before the Bill

Quebec does not need to copy PJM to learn from its auction. Before every promise of megawatts, it must demand a public response regarding marginal cost, necessary infrastructure, sustainable jobs, and exit clauses. Otherwise, today’s discount will become tomorrow’s shortage.

This debate calls for neither closure nor naivety. It calls for prioritization. A megawatt allocated to a campus is no longer available for a factory, a neighborhood, a switch to electric heating, or export. Making choices is legitimate. Hiding those choices is not.

Scarcity does not disappear just because we call it economic development.

The Auction’s Verdict

Neither panic nor naivety

The July auction does not prove that every data center impoverishes every household. Nor does it allow us to definitively isolate the 6.3 billion from all other factors in the grid. To claim otherwise would turn a solid warning into a fragile caricature.

It does, however, prove that demand is growing faster than supply, that prices have hit the ceiling, that reserved capacity remains below standard, and that the independent auditor attributes a major portion of the additional cost to data centers. That is more than enough to take action.

Progress That Pays Its Own Way

I want an America capable of winning the race for artificial intelligence. I don’t want it to win by making those who have no seat at the table pay for the heat generated by its machines. Progress deserves better than a disguised subsidy.

Let’s accelerate meaningful connections. Let’s build out the infrastructure. Let’s modernize the lines. Let’s protect low-income households. And above all, let’s make the major beneficiaries pay the true cost of their arrival.

$6.3 billion. The figure isn’t yet an individual bill. It’s already a collective issue.

Signed, Maxime Marquette, columnist

Columnist’s Transparency Box

Editorial Stance

I am not a journalist, but a columnist and analyst. My expertise lies in observing and analyzing the geopolitical, economic, and strategic dynamics that shape our world. My work consists of dissecting political strategies, understanding global economic trends, contextualizing the decisions of international actors, and offering analytical perspectives on the transformations that are redefining our societies.

I do not claim to possess the dispassionate objectivity of traditional journalism, which is limited to factual reporting. I aim for analytical clarity, rigorous interpretation, and a deep understanding of the complex issues that affect us all. My role is to make sense of the facts, place them within their historical and strategic context, and offer a critical analysis of events.

Methodology and Sources

This text respects the fundamental distinction between verified facts and interpretive analysis. The methodological rule is consistent: factual information is published only if it is supported by a verifiable source, and the sources actually used in this article are listed under “Sources,” never here.

Categories of primary sources used by the publication, when applicable: official press releases from governments and international institutions, public statements by political leaders, reports from intergovernmental organizations, and dispatches from recognized international news agencies.

Types of secondary sources: specialized publications, internationally recognized news media, analyses from established research institutions, and reports from sector-specific organizations.

When an article cites statistical, economic, or geopolitical data, it comes from data-producing institutions (intergovernmental organizations, central banks, national statistical institutes), and the specific institution is listed under “Sources.”

Nature of the Analysis

The analyses, interpretations, and perspectives presented in the analytical sections of this article constitute a critical and contextual synthesis based on available information, observed trends, and expert commentary cited in the sources consulted.

My role is to interpret these facts, contextualize them within the framework of contemporary geopolitical and economic dynamics, and give them coherent meaning within the broader narrative of the transformations shaping our era. These analyses reflect expertise developed through continuous observation of international affairs and an understanding of the strategic mechanisms that drive global actors.

This article describes a situation documented as of its publication date, not a prediction: subsequent developments may alter these perspectives. No updates are promised in advance; when an article is corrected or supplemented, the change is dated within the text.

ANALYSIS: PJM, Artificial Intelligence, and the Bill We Pass On to Our Neighbors