We didn’t see the fuse. At least I didn’t. For most of this cycle, the story was simple: artificial intelligence needs electricity, Bitcoin mining needs electricity, and the grid will somehow find a way. Then Constellation Energy’s CEO stepped up and said the quiet part out loud: existing power plants are the bedrock for data centers. No waiting for small modular reactors. No waiting for green hydrogen. No waiting for a grid rebuilt from scratch. Just use the plants already there. Immediate. Reliable. Available.
The market heard one thing: nuclear is back. I heard something else: a power producer with a balance sheet full of legacy machines is trying to convert regulatory urgency into pricing power. This is not a story about energy. It is a story about yield migration. And if you are a crypto native, you are not a spectator. You are the counterparty.
For two decades, I have watched narratives harden into orthodoxy. The same way DeFi Summer taught us that yield farming was a social contract, and the same way the NFT boom taught us that status signaling can move 10,000 ETH worth of volume, the current “power is scarce” narrative is being manufactured in real time. The Constellation CEO is not just describing a physical fact. He is building a framework. And every framework has a beneficiary.
This article is not an attack on nuclear power. It is an autopsy of a sales pitch. Let me walk you through why the phrase “existing power plants are bedrock” is technically true, strategically convenient, and possibly a trap for everyone who signs a 20-year contract based on it.
The Scarcity Story Is Real
Before we get to the contrarian layers, we have to acknowledge the foundation. The scarcity is not fabricated. US data centers currently consume roughly 4% of all electricity consumed in the country. By 2030, depending on which consultant you quote, that share will double to somewhere between 8% and 10%. That is not a linear growth curve. That is a hockey stick.
The problem is on the supply side. New gas plants take years to permit. New nuclear plants take decades, if they happen at all. Solar and wind are cheap but intermittent, and in the United States, the average interconnection wait for new renewable capacity is now more than four years. Transformer deliveries, which used to take under a year, are stretching to two, three, or four years. The grid is not just tight. It is physically walled off.
So what do you do when demand is exploding and new supply is trapped in queues? You start looking at the assets that already exist. That is where Constellation Energy comes in.
Constellation is the largest owner of nuclear power plants in the United States. It also owns natural gas generation. It has a fleet that can run 24/7, independent of sunlight and wind. In a world where “immediate and reliable” are the only two words that matter, a nuclear fleet is the equivalent of a printing press for electrons.
And the numbers back this up. The PJM capacity auction for 2025/2026 cleared at roughly $268.9 per megawatt-day, up from $28.9 in the same auction a year earlier. That is not a mild correction. That is a nine-fold spike. It is the price signal of a market that has realized there is not enough firm capacity to satisfy the AI buildout. Constellation, as one of the largest owners of firm capacity, is standing in the middle of a profit avalanche.
This is why the CEO’s statement matters. He is not just giving an opinion. He is providing the narrative cover for the next decade of electricity pricing.
The Ledger’s Silence: What the CEO Does Not Say
In my line of work, I always look for what is missing from the transcript. The CEO says “existing power plants are bedrock.” He does not say “new nuclear will save us.” He does not say “batteries will save us.” He does not say “solar and wind are partners.” He says: use what is already here.
That is a very deliberate choice.
Constellation owns the existing nuclear fleet. It has already recovered the construction cost of those plants, or at least accounting fiction has pushed those costs deep into depreciation schedules. The marginal cost of operating a nuclear plant is low, something like $30 to $60 per megawatt-hour for the older plants. But in recent negotiated agreements with hyperscalers, the effective price has been reported at much higher levels. The Microsoft deal to resurrect Three Mile Island is widely believed to involve a power purchase agreement in the range of $115 per megawatt-hour. That is a massive spread over marginal cost. That is not energy pricing. That is rent extraction disguised as reliability.
Now, I am not saying Constellation does not deserve a return. It took on a lot of risk to keep these plants alive. But the narrative being packaged for the public is one of national necessity, while the underlying mechanism is one of asset scarcity. The phrase “existing power plants are bedrock” is designed to make the scarcity of new supply feel permanent. If you believe that, you will sign any contract.
I have been fooled by a confident narrative before. In 2018, I published a bullish thesis on Raptor Protocol after reverse-engineering its smart contracts for about 40 hours. I was convinced the yield strategy was revolutionary. A week later, the protocol was drained by a reentrancy bug. I learned something that has stuck with me ever since: code is law, but humans write the bugs. And when someone tells you an asset is “the bedrock,” the real question is, whose balance sheet is underneath that bedrock?
The False Binary: “Old Power vs. New Batteries”
One of the most effective tricks in narrative construction is to force a binary choice. The Constellation CEO’s framing implies that the only alternatives are existing power plants or some speculative future technology. If you accept that binary, the answer is obvious: existing plants win.
But reality is not binary. The actual future of data center power will not be a choice between nuclear and solar, or between batteries and gas. It will be a hybrid system. You will need nuclear or gas for baseload. You will need batteries for milliseconds-to-hours of response. You will need demand response to shave the sharpest peaks. You will need renewables to lower the average cost of the energy mix. And you will need all of these technologies working together, through software, to keep the system stable.
Why does the CEO not describe that vision? Because a hybrid system does not require a single dominant asset provider. It requires ecosystem thinking. It requires multiple vendors, multiple revenue streams, and less pricing power for any one player. Constellation wants to be the anchor. The “existing power plants” narrative is how it anchors the market to its balance sheet.
Let’s look at the storage option seriously. Battery costs have collapsed. LFP cells now cost something like $0.50 to $0.80 per kilowatt-hour on a levelized basis, depending on the market and the application. But batteries are not equivalent to a power plant. They provide four hours, maybe six hours, of stored energy. If you have a 500 megawatt data center and the grid goes down for a week, you are not building a battery park. You are building a second data center full of batteries. That is not a solution; that is a physics problem.
Long-duration storage, like flow batteries or compressed air, is still in the early commercial stage. The technology is promising, but it is not currently deployable at the scale and speed that data centers need. So yes, on a purely technical basis, the CEO is correct: batteries cannot yet replace a baseload plant.
But notice what this does to the argument. It conflates “batteries alone” with “renewables plus storage.” It implies that if the battery answer fails, then the only answer is old power. That is a false binary. The most likely deployment is a gas peaker plant next to a solar field next to a four-hour battery, all managed by a virtual power plant operator. That is not a threat to Constellation. It is a different architecture. And the architecture matters because it determines who captures the yield.
Sentiment Is a Shifting Tide, Not a Solid Ground
I keep returning to the phrase “immediate and reliable.” It sounds like a technical requirement. But it is also a sentiment weapon. “Immediate” weaponizes time. “Reliable” weaponizes fear.
When a hyperscaler hears “immediate,” it thinks: we cannot wait for a new nuclear plant to be permitted and built in ten years. When it hears “reliable,” it thinks: we cannot let our uptime drop below 99.99%. The combination creates a willingness to pay almost any price. That willingness is exactly what the old-power narrative needs.
Sentiment is a shifting tide, not a solid ground. I have seen it in crypto markets a hundred times. When the market is afraid, it clings to the asset that appears safest. In 2020, it was centralized exchanges. In 2022, it was US Treasuries. In 2025, it is existing nuclear plants in America. But the sentiment of safety is not the same as safety. It is a premium that someone is being paid to adopt a particular story.
The actual risk picture is more complicated. An existing nuclear plant is “reliable” until a transformer fire forces a derate. It is “reliable” until a cooling water restriction cuts output during a heat wave. It is “reliable” until the fuel supply chain, still partially dependent on Russian enriched uranium for a portion of US inventory, gets squeezed by geopolitical forces. The plant is not a bedrock. It is a machine. All machines break.
The word “existing” is itself a warning. Existing plants are aging. Many were designed for a world where the electric grid was predictable and demand growth was low. Now they are being asked to run at maximum capacity, serving a load profile that is spikier and more aggressive than anything they were designed for. That shortens the remaining life of the fleet and increases the probability of forced outages. In other words, the exact assets the CEO is hyping are the ones most likely to create the next reliability crisis.
The Rise of the Gigawatt Buyer
There is another hidden player in this story: the technology companies themselves. Microsoft, Google, Amazon, and Meta have become some of the largest electricity buyers in the world. They are not passive ratepayers. They are active financial engineers. They sign PPAs, they buy capacity, they invest in startups, and they are increasingly willing to do whatever it takes to secure power.
This is where the blockchain connection gets sharp. For years, Bitcoin miners were the villains of the energy narrative. They were accused of eating the grid and destroying the planet. But Bitcoin mining is actually one of the most flexible loads ever created. A Bitcoin miner can shut down in under a second. It can curtail, shift, and even sell power back to the grid during peak hours. Many mining operations now function as demand-response resources. They are not the grid’s enemy; they are the grid’s shock absorber.
Data centers are different. AI inference and training cannot simply stop when the price spikes. You cannot pause a large language model training run for two hours and resume it later without massive consequences. Data centers need steady, high-quality, persistent power. They are the new inflexible load. And they are willing to pay a huge premium.
What does that mean for Constellation? It means the CEO no longer needs to market electricity to homeowners or industrials. He needs to market security to trillion-dollar tech companies. And the best way to do that is to make them afraid of their own shadow. “Existing power plants are bedrock” is a sentence designed for people who have not slept well in months.
The parallel to crypto is almost too obvious. In every bull run, there is a narrative that justifies high prices. “Layer 2s are the solution.” “Institutions are coming.” “This time it is different.” Every bull run is a myth waiting to be debunked. The current bull run in the power market has its own myth: the myth of the unbreakable legacy plant. I am not saying it will collapse tomorrow. I am saying the myth is real, and it has a function. It is there to transfer wealth from people who must have power to people who already own it.
The Contract Trap
Now we reach the part that should make every CFO, every crypto miner, and every infrastructure investor nervous: the contract trap.
When the CEO says “existing power plants are bedrock,” he is making a claim about the future. He is saying that these assets will continue to generate, that fuel will be available, that the regulatory climate will remain favorable, and that the price of electricity will stay high enough to justify the long-term contract. But he is not bearing all the risk. The counterparty is.
Consider the structure of a typical PPA for nuclear power. The buyer commits to a fixed or floor price for power for 20 years. That protects the seller against price declines. The seller gets a guaranteed revenue stream. The buyer gets reliability. But what if the plant gets retired earlier than expected? What if the PPA is structured around a capacity credit that is reduced by regulators? What if the buyer’s actual demand falls because AI costs decline and data center buildouts slow? The buyer is still on the hook.
This is exactly the pattern I saw in crypto lending. Yield is the bait, liquidity is the trap. In decentralized finance, protocols lure users with high yields, then pull the liquidity from underneath them. The same principle applies to old-power contracts. The “yield” is the promise of stable electricity. The “liquidity” is the flexibility to escape the contract when the market turns. And once you sign a 20-year power purchase agreement, that flexibility is gone.
I am not suggesting that every nuclear PPA is a bad deal. Some are genuinely necessary to restart plants that would otherwise retire. The restart of Three Mile Island is a real industrial achievement. But the narrative that surrounds it—that old power is bedrock, that nothing else can work, that the only choice is between legacy plants and chaos—is designed to reduce the buyer’s perceived options. And when options disappear, price discipline disappears.
Cryptocurrency miners understand this better than anyone. They were the first to be squeezed by power prices, curtailment, and grid fees. They know that a mining rig is only worth money if the electricity cost stays below the network hashrate. When the power price spikes, miners shut down. That flexibility is the entire business model. If crypto miners lose that flexibility—if they sign long-term contracts that force them to pay during a bear market—they become just another fixed-cost load, with none of the upsides.
The next wave of crypto and AI convergence will be about autonomous agents, micro-payments, and machine-to-machine transactions. But none of that matter if the physical power layer is dominated by a handful of incumbents who can extract rent. In the ledger’s silence, the true story whispers: the real value is not the electrons. It is the optionality to avoid becoming a permanent victim of someone else’s scarcity narrative.
The Contrarian Blind Spot: What “Existing” Actually Costs
Let me play the contrarian card again, because this is the part that mainstream coverage keeps missing.
The conventional reading of the Constellation CEO’s statement is: old plants are good, new tech is bad. The contrarian reading is: old plants are a liability disguised as an asset.
First, there is the fuel issue. Nuclear fuel is cheap per unit of energy, but the supply chain is concentrated and vulnerable. The US still imports a meaningful share of its enriched uranium from Russia. The 2028 ban on Russian imports is real, but the transition is not smooth. A fuel supply shock is not impossible. If Constellation has a fleet of reactors that need fuel, and the fuel supply is geopolitically fragile, then “existing” is not strength. It is concentration risk.
Second, there is the aging infrastructure issue. Many of these plants have been operating for forty or fifty years. They have been relicensed and uprated, but the physical components still wear out. A single major component failure can shut a plant down for months. The expected outage costs are not always captured in the PPA price. When a plant goes offline, the buyer has to go back to the spot market, where prices are even more volatile. The “bedrock” disappears exactly when you need it most.
Third, there is the regulatory risk. Nuclear power is heavily regulated. A single change in safety rules, cooling water standards, or spent fuel storage policy can alter the economics of a plant overnight. The CEO is betting that the current pro-nuclear, reliability-first political mood will persist for decades. That is a political bet, not an engineering fact.
Fourth, there is the missed opportunity cost. If data center operators lock themselves into 20-year contracts with existing nuclear plants, they are simultaneously locking themselves out of future technological improvements. Hydrogen fuel cells, advanced geothermal, small modular reactors, and long-duration storage are all improving quickly. Ten years from now, the cost of a firm power system might be much lower than today. But if you have already signed a contract at today’s scarcity price, you will miss that opportunity.
So the contrarian position is not “renewables will save us.” The contrarian position is “legacy power is not a safe harbor; it is a toll booth.” And the tolls will only grow.
The Real Battle: Power as the New Asset Class
If I could rewrite the headline, I would call this story “Power Is the New Crypto.” The energy market is beginning to function like a crypto market: sentiment-driven, narrative-manipulated, and dominated by a few players who control the supply of a scarce asset. The same games that played out in DeFi—liquidity mining, yield farming, Ponzi-like spirals—are now playing out in the power grid.
Think about it. PJM capacity prices spiked nine-fold in a year. That is not an energy market. That is a speculative bubble. It is driven by a sudden, fundamental demand shift, yes. But the price level is amplified by fear, by uncertainty, and by the unshakeable belief that there is not enough physical power to go around. In that sense, the grid has become a blockchain: an immutable ledger of physical constraints, with massive price swings around the edges.
The biggest difference is that the power market cannot be forked. You cannot just spin up a new grid in a weekend. The response time is measured in decades. That makes incumbents even more powerful. It also means that when the bubble pops—and I repeat, when, not if—the consequences will be brutal for anyone who bought the narrative at the top.
What would pop the bubble? A recession that slows AI investment. A breakthrough in energy efficiency that cuts data center demand. A dramatic increase in flexible load management using smart software. A wave of small modular reactor deployments that actually arrives on schedule. Any of these would reduce the scarcity premium. And when it happens, the long-term contracts signed at today’s prices will look like financial weapons of mass deconstruction.
The same dynamic is playing out in crypto mining. Hashprice has fallen because more miners keep joining the network, even during a bear market. Miners who locked in high power costs during the bull run are now forced to sell their coins at a loss to pay the electricity bill. The ones who survived are the ones who retained flexibility, who built on gas flares, or who negotiated interruptible load tariffs. The ones who died are the ones who treated power as a fixed cost rather than a tradable risk.
That is the lesson for this next phase. Power is not a utility. It is a derivative. It has a spot price, a forward curve, an options market, and a basis. The people who thrive will be the ones who think of power like a trader, not a consumer. The people who get trapped will be the ones who just want to “set it and forget it.”
The False Promise of “Immediate”
“Immediate” is the other word that deserves scrutiny. The CEO says the sector needs “immediate, reliable” energy. But “immediate” is not a quality of the power plant. It is a quality of the contract. Existing power plants are not inherently immediate; they are already connected. That is a huge advantage.
But “immediate” can also mean “expedient.” It can mean “we do not want to wait for a better solution.” In the actual data center market, the rush to secure power is creating a preference for whatever can be connected fastest. That gives an unfair advantage to any asset that is already connected, regardless of its long-term viability.
This is where the crypto analogy gets personal. I have seen projects raise millions of dollars by selling “immediacy.” The narrative is always the same: the network is congested, the fees are high, and you need to pay a premium to get your transaction through now. In reality, there is almost always a cheaper path that requires a little more patience. The same is true for power. The cheapest path might not be the fastest. The fastest path might not be the most resilient.
I am not saying Constellation is lying. I am saying the word “immediate” is doing an enormous amount of work in the sentence. It is a heuristic that favors the seller.
The Hybrid Future Is Not a Fantasy
Let me sketch the alternative future that is actually plausible.
A data center campus in Ohio signs a 15-year deal with a gas-fired plant for baseload power. It also signs a variable PPA with a nearby solar project. It installs a 4-hour battery system on-site to ride through short outages and to arbitrage the wholesale market. It uses software to shift non-critical workloads to off-peak hours. It maintains a small hydrogen fuel cell backup for emergency reliability, supplied by a green hydrogen project that is slowly scaling up. Ten years later, as solar costs keep falling and storage lasts longer, the campus shifts more of its load to renewable-plus-storage and uses the gas plant only for the darkest, calmest winter nights.
Is that fantasy? No. Every component exists today. The integration is hard, but the engineering is known. What is missing is not technology. It is the willingness to design an energy system rather than simply buy a commodity.
The Constellation CEO’s message is seductive because it offers a shortcut. It says: stop worrying, we already have the power plants. But that shortcut is only a shortcut for the seller. For the buyer, it can become a one-way door to a world where the energy mix is locked in, the price is fixed, and the flexibility is gone.
I have watched too many crypto projects get trapped by their own narratives. In 2021, every “Ethereum killer” promised immediate scalability. Most of them are dead. The ones that survived were the ones that kept building through the bear market. The same principle applies to power. The plants that will be most valuable in 2035 are not necessarily the ones that are most politically convenient in 2025. They are the ones that can adapt to a changing fuel mix, a changing regulatory environment, and a changing load profile.
The Takeaway: Don’t Be the Exit Liquidity
So what should a crypto-native reader take away from this?
First, power is the bottleneck. The AI economy is going to need enormous amounts of electricity. Crypto mining, if it survives, is going to need its own supply. The price of that supply is going to be volatile, and the survivors will be those who treat it as an active risk, not a passive cost.
Second, narratives are expensive. The “existing power plants are bedrock” narrative will push many companies into long-term contracts at today’s scarcity prices. Some of those contracts will be rational. Many will not. The key is to separate the physical quality of the asset from the narrative quality of the pitch.
Third, flexibility is the only edge. In a bear market, survival matters more than gains. The protocols that survive high volatility are the ones with enough liquidity to weather the storm. The same is true for energy buyers. If you sign a contract that removes all your flexibility, you become the exit liquidity of someone else’s trade.
I have been an analyst for a long time. I have been wrong, and I have been right. The one thing I trust is the pattern. Every bull run is a myth waiting to be debunked. The myth here is that existing power plants are a permanent bedrock. They are not. They are a finite resource, a strategic asset, and an economic weapon. And the people who sell that myth are already on the other side of the trade.
We didn’t see the fuse. But it is lit. The only question is whether you will be holding the matches.
In the ledger’s silence, the true story whispers: the next great trade is not a token. It is the right to stay flexible in a world that will do anything to make you sign, lock in, and obey. The night is long. The grid is old. The prices are climbing. And the only guarantee is that somebody will be left holding the power bill.