America's largest grid just came up short on its own safety cushion

Climate & Energy 4 min 80 sources

A once-a-year auction that buys the power grid's spare capacity cleared far below target — a quiet number that tells you where electricity bills, data centers, and the energy transition all collide.

Key takeaways

America's largest grid operator held its once-a-year auction to buy spare "insurance" capacity and came up billions of dollars and gigawatts short of its own safety target.
Demand from data centers is climbing far faster than new power plants can be built, and even the cheapest new supply — solar and batteries — is getting pricier or arriving too slowly to close the gap.
The squeeze is already showing up in electricity bills, which are rising faster than inflation, even as solar quietly becomes Europe's biggest single source of power.

The most important climate-and-energy number this week wasn’t a temperature. It was the result of an auction most people have never heard of — and it says a lot about where your electricity bill is heading.

The grid’s safety cushion came up short

PJM, the largest grid operator in the United States, runs a market that pays power plants years in advance to promise they’ll be available on the worst day of the year. It’s insurance for the grid. This week’s auction, for the 2028–2029 delivery year, hit its $325 per megawatt-day price cap across the whole region [34].

Even at that capped price, the auction came up roughly 6.8 gigawatts short of PJM’s target — a 20% cushion of spare capacity above expected peak demand [34]. That’s a bigger gap than last time, which was itself the first shortfall in the auction’s history [34]. And it pulled in only about 500 megawatts of genuinely new supply, down from 774 the auction before [34]. PJM serves 67 million people across 13 states and Washington, DC [71].

The price cap masked how tight things really are. Without a temporary lid called a “price collar,” PJM said the auction would have cleared near $555 per megawatt-day region-wide — and $777 in the northern Illinois zone around Chicago [34]. The bill for the year would have been $29.7 billion instead of $16.4 billion [34].

Federal regulators noticed. “These numbers compound the alarm bells for a call to action in PJM,” said Laura Swett, who chairs the Federal Energy Regulatory Commission [71]. “Am I surprised that PJM failed to deliver? No, I am not” [71].

Why the cushion is so hard to refill

The gap is really a race between two things moving at very different speeds: demand, which is climbing fast, and new supply, which arrives slowly.

Bank of America analysts estimate the US will need more than 230 gigawatts of new generating capacity over the next five years, but regulated utilities are on track to add only about 93 — a shortfall of more than 100 gigawatts [25]. Data centers alone, the warehouses of computers behind AI and the cloud, could add roughly 125 gigawatts of demand, pushing electricity growth to about 4% a year [25]. Large gas turbines, the fastest way to add firm power, are largely sold out through 2030 [25].

New supply is also getting more expensive to build. Renewables are still the cheapest source of new electricity, but their lifetime cost is rising for the first time in years, according to a closely watched Lazard analysis — squeezed by tariffs, higher interest rates, and strained supply chains [44]. And developers say the auction’s $325 cap is nowhere near enough to make a new plant worth building [34]. The market is asking for more capacity while capping what it will pay for it.

The bill, and where the clean supply is coming from

For most PJM households, this particular auction won’t jolt the bill much — it cleared close to last year’s price, and takes effect in mid-2028 [34]. But the direction is unmistakable. A 10-megawatt industrial customer’s monthly capacity charge is set to jump from around $6,000 in 2024 to roughly $70,000 in 2028, one energy strategist estimated [34]. And US retail electricity rates already rose 2.6% above inflation from 2024 to 2025; residential prices are up 33% since 2019 [76].

The one bright spot on supply is solar, which is scaling faster than anything else. It has now become Europe’s single biggest source of electricity, capping years of roughly 20%-a-year growth [5]. The EU added 65 gigawatts of solar in 2025 alone; Spain hit 34% of its power from solar in June, Germany 36% [5].

The missing piece: cheap storage

Solar’s weakness is obvious — it stops at sundown, exactly when demand peaks. Making it count toward the grid’s cushion means storing it, and that’s where a quieter race is running.

Beyond lithium-ion batteries, developers are testing stranger chemistries: liquid air, molten salt, and “flow batteries” that store energy in tanks of liquid rather than solid cells [18]. Flow batteries can charge and discharge almost indefinitely and skip the scramble for lithium, but they’ve been held back by cost and by results that were hard to reproduce between labs [28].

This week’s under-covered story is a small fix for that. A PhD researcher designed a standard flow-battery test cell — about ten parts, roughly £74 — and, rather than sell it, gave the design away free to the global research community with an “Ikea-style” manual so scientists everywhere test on identical hardware [28]. It won’t power a city. But cheaper, comparable experiments are how a lab curiosity becomes a grid resource — the kind of spare capacity PJM just couldn’t buy.

Meanwhile the glut that made solar cheap is punishing its makers: Chinese module manufacturers are forecasting heavy losses as oversupply drags on [69]. Cheap panels for the world, red ink for the factories that build them.

02 · Lesson · why it matters

Why a grid is built for its worst hour, not its average

A system that has to survive its worst moment carries spare capacity that looks like waste — until the day the cushion runs thin, when its price stops climbing and starts leaping.

The number that wasn’t a temperature

The biggest energy story this week was a shortfall in an auction. Not a blackout, not a heatwave — an auction, three years ahead of the electricity it’s about. America’s largest grid operator tried to buy a cushion of spare power for the summer of 2028, and came up billions of dollars short, even after the price slammed into its ceiling.

To see why a missed auction matters more than a hot afternoon, you have to understand a strange fact about the grid: it is not built for the electricity we normally use. It is built for one hour a year we hope never comes.

A machine sized for its worst moment

Think about the electricity you use across a year. Almost all of it sits well below your peak. Then comes a still, brutal evening in July: the air conditioners all run at once, nobody’s solar is generating, and demand spikes to its highest point of the year. The grid has to hold up in that hour — or the lights go out for millions.

So planners don’t size the system for the average. They size it for the worst hour, and then they add a cushion on top: build about 20% more capacity than even that peak is expected to need. That cushion is spare plants that mostly sit idle. For thousands of hours a year it looks like pure waste — capacity you’re paying for and barely touching.

It isn’t waste. It’s the margin between a comfortable summer and a catastrophe. And the whole business of running a grid is the business of carrying that margin — deciding how thick it should be, and paying to keep it there.

Why the price leaps instead of climbs

Here’s the part that surprises people. When the cushion gets thin, the cost of it doesn’t rise gently in proportion. It jumps.

The grid buys its cushion at auction, and the price is set by the last, most desperate megawatt needed to hit the target. When there’s plenty of spare capacity, that last megawatt is cheap. But as the system approaches the edge — as supply just barely covers what’s required — that final megawatt becomes precious, and its price runs away. This week’s auction hit its price cap. Regulators noted that without a temporary lid, it would have cleared far higher still, nearly doubling the bill.

That is the signature of scarcity near a threshold. Far from the edge, a shortfall of a few percent costs a little. Close to the edge, the same few percent costs everything the market will bear. The line between “fine” and “frightening” is not a gentle slope. It is a cliff, and the price is what falls off it first.

The cushion is bought years early, and for good reason

Notice the timing: this was an auction for 2028, held now. The grid tries to lock in its cushion three years ahead, because the thing it’s buying — a new power plant, a battery farm, a transmission line — takes years to build.

That lead time is the trap the grid is caught in. Demand can move in months. A cluster of data centers, the buildings full of computers behind AI, can be announced and drawing power faster than any plant can be poured and wired. So the cushion thins from the demand side quickly, and refills from the supply side slowly. An auction coming up short today is really a message about a race that was lost quietly, over years — supply that was never started early enough to be ready.

Who is standing inside the margin

It’s tempting to read all this as a story about one grid operator and its regulators. It isn’t. Everyone inside that region — 67 million people — lives on the far side of that cushion, whether they know the auction happened or not. The price of the margin flows into ordinary bills, which is why electricity has been rising faster than inflation. You are paying, right now, for spare plants you will likely never notice, so that one evening in some future July doesn’t become a disaster.

And the pattern reaches far past the grid. A hospital keeps empty beds for the night a bus crashes. A water utility sizes its reservoir for the drought decade, not the wet one. A household keeps a savings buffer for the month everything breaks at once. Every one of these is a cushion — capacity that looks like slack in good times and turns out to be survival in bad ones. And every one of them has the same cruel arithmetic: run the margin thin to save money, and most years you win, until the year you don’t, when the cost of being caught short leaps instead of climbs.

The whole, held loosely

What looks like waste is often a system carrying its own insurance. The spare that a lean eye wants to trim is the exact thing standing between an ordinary day and a bad one — and its value is invisible precisely because it usually isn’t needed. The people who cut the cushion to look efficient, and the people who pay to keep it, are both partly right; the argument between them is never settled from inside a single quiet year.

The grid, the hospital, the reservoir, the savings account — each is a decision made now against a worst hour we can’t see coming, priced by a market that punishes lateness without mercy. It’s worth remembering how much of the comfort we take for granted rests on a cushion someone chose to carry, at a cost, on our behalf. And how little any one of us, looking only at our own calm afternoon, can tell about how thin that margin has quietly become.

03 · Lab · your turn

Sizing the Cushion

Rehearse how much spare grid capacity to carry, and feel why running the cushion thin stays cheap for years then leaps to the cap in the bad one.

04 · Hope · carry this

A thin cushion today is not a permanent one. This same week solar became Europe's biggest source of power, and a young researcher gave his battery design away for free — the tools to refill the margin are arriving, and the people building them still choose to share.