Submarine Reactors, SPAC Money, and the Power Shortage the Market Is Misreading

Trust no one. Verify everything. That is the first rule of infrastructure finance, and it is the only discipline that can survive a meeting with a decommissioned submarine. A brief crossed my desk this week, published in Crypto Briefing of all places. Not a crypto story. Not exactly an energy story either. More like a ghost from the future wearing a maritime uniform. HGP Intelligent Energy, the vehicle formerly described in connection with Holtec Government Services, is reportedly exploring a SPAC listing at a valuation near $1 billion. Its pitch: take retired naval reactors, pressurized water reactors originally built to push submarines through cold Atlantic water, and re-license them as land-based power plants for AI data centers. Navy reactors. AI workloads. SPAC capital. I have seen each of those ingredients before, and none of them spoils quickly. But the mixture is new, and the timing tells me more than the technology does. Strip away the hull metal and the marketing deck, and the core finding is not that HGP will succeed. The core finding is that a market signal has formed around a structural marriage between AI compute demand and clean baseload power. Even if this particular deal drowns in the regulatory shallows, the demand signal it is trying to ride will not die with it. Noise is cheap. Signal is rare. Let me be precise about what the source actually says, because most commentary will be less precise. The article describes a plan by HGP Intelligent Energy to use decommissioned naval reactors for AI data center power, with an SPAC transaction intended to bring the project to public markets at a $1 billion valuation. It openly admits that regulatory obstacles and economic feasibility remain critical challenges. That admission is the only honest sentence in the entire pitch. The rest is a roadmap disguised as a press release. Now let me put this inside a frame that makes sense. In 2020, I spent months inside the MakerDAO governance swamp, watching idealistic token models bend under whale pressure. In 2021, I watched artists sell their soulbound identities the moment the market offered a flip price. I have learned to read the gap between the story an asset tells and the balance sheet it keeps. A retired nuclear reactor is an asset with a story. The question is whether the balance sheet can carry the weight. And the balance sheet of the American grid is exactly where this story belongs. The causal chain underneath the HGP narrative has nothing to do with submarines and everything to do with load curves. AI inference and training clusters do not behave like households. They do not sleep. They do not shut down for summer holidays. A hyperscale data center runs at utilization rates north of 80 percent, sometimes above 90 percent, and it demands power with a reliability profile that wind and solar cannot honestly promise without heroic storage assumptions. I have watched this demand curve bend over three years of direct observation. In 2023, the conventional estimate for U.S. data center electricity consumption was somewhere near 20 gigawatts. By 2030, depending on which model you trust, that number is expected to land somewhere between 60 and 100 gigawatts. That is not a linear extrapolation. That is a hockey stick. And hockey sticks break grids. Every regional transmission operator in the United States is now living inside that breakage. PJM, ERCOT, and others are reporting interconnection queues measured in years, not quarters. Grid operators are telling data center developers that a new connection can take three to seven years. For a technology sector that plans its capacity in eighteen-month cycles, seven years is geological time. That timeline gap is the entire reason decommissioned naval reactors are suddenly interesting. Wind and solar remain the cheapest electrons on earth, down to roughly $20 to $40 per megawatt-hour for the generation itself. Add enough storage to make a 24/7 data center run on intermittency, and the all-in cost lands between $60 and $100 per megawatt-hour while still failing the reliability test. A multi-day storage requirement, the kind that would be necessary to carry a data center through a winter week of low wind and heavy cloud, is economically absurd at current battery prices. Natural gas combined-cycle plants are more dependable, around $40 to $60 per megawatt-hour, but they emit carbon, and the hyperscalers have painted themselves into net-zero corners. Then there is nuclear. Zero-carbon, high-capacity-factor, fuel cost that is almost irrelevant to the total cost structure. The catch has always been capital cost and construction time. A large new plant is a decade-long, multi-billion-dollar prayer. But what if the plant already exists? What if the reactor was built decades ago, maintained by the world's most disciplined bureaucracy, and is now sitting in a nuclear submarine that no longer patrols? That is the HGP shortcut. It is a simple thought, and like many simple thoughts, it contains a hidden trap. The trap is that a naval reactor is not a plug-and-play generator. The U.S. Navy's pressurized water reactor designs are mature, arguably the most reliable fission machines ever operated, with millions of accumulated safe operating hours. That is the strongest argument in HGP's favor. But moving from a submarine hull to a commercial site is not like moving a diesel generator from a garage to a construction site. It is a conversion of one regulatory species into another. Naval propulsion has historically sat under the Navy Nuclear Propulsion Program, or NNPP, which is not the same jurisdiction as the civilian licensing framework of the Nuclear Regulatory Commission. A retired Navy reactor has spent its entire service life inside a military classification envelope. Its core was designed for compactness, for shock resistance, for the peculiar demands of undersea warfare, not for the civilian safety culture built around large light-water plants with massive containment structures and low-enriched fuel assemblies. Here is the detail almost every summary will omit. Many American naval propulsion cores have used highly enriched uranium, material that is effectively one short step from weapons-grade. Under the civilian NRC framework, commercial power reactors operate on low-enriched uranium, capped well below 20 percent U-235. A decommissioned submarine core may exceed that threshold by a wide margin. Bringing such a core into commercial service, or even removing it and replacing it with a fresh low-enriched core, is not merely an engineering challenge. It is a nuclear nonproliferation challenge of the first order. International safeguards exist precisely to prevent the normalization of highly enriched uranium in commerce. The International Atomic Energy Agency would treat the civilian use of naval HEU as a test case. Other countries, including Russia and China, would be handed a rhetorical gift. The United States has spent decades arguing that HEU should be minimized in civilian applications. A commercial data center powered by a submarine core with weapons-grade fuel would contradict that entire policy architecture. That contradiction will not be resolved by a PowerPoint slide. And the fuel is not the only uncomfortable inheritance. The United States has historically preferred to dispose of decommissioned naval reactor cores by encapsulating them intact, often in burial vaults, rather than cutting them open and shipping the pieces across the country. The Navy treats spent naval cores as sensitive material with a high security perimeter. Transferring that material to a private company, and then to a commercial site, would require the government to reverse a long-standing institutional habit. Government institutions are slow to walk away from their own precedents. Let us also be honest about the reactor itself, because a $1 billion valuation deserves a technical audit. A single decommissioned submarine or cruiser reactor, of the S8G family that has circulated in industry reporting around this project, is a small module. The thermal power is on the order of 100 to 300 megawatts, with electric output plausibly in the 30 to 100 megawatt range. Some estimates for retrofitted naval modules land near 50 to 80 megawatts electric. That is a meaningful chunk of power for a single data center campus, but it is not a grid-scale savior. This is a community-scale machine, not a colossus. The supply of these machines is finite. Only a limited number of retired naval reactors exist, and not all are economically recoverable. Even under generously optimistic assumptions, the total deployable capacity from this niche by 2035 is probably no more than two to three gigawatts. Meanwhile, AI data centers are expected to add tens of gigawatts of new demand over that same period. Two to three gigawatts against tens of gigawatts is a rounding error. It is a niche market, not a revolution. And when a SPAC pitch calls a niche a revolution, my audit instinct begins to itch. That itch is grounded in experience. In 2017, during the ICO mania, I audited fifteen early Ethereum protocols and found that most of them were marketing dressed as mathematics. I wrote a long essay called Math Over Hype because I was tired of watching investors confuse narrative volatility with technical integrity. The same confusion is now appearing in the energy transition. A real nuclear asset is being wrapped in the same narrative mechanics that produced a thousand worthless tokens. Let me compare HGP against the companies it will actually be measured against. Microsoft has signed a twenty-year power purchase agreement with Constellation Energy to restart Unit 1 at Three Mile Island, an 835-megawatt plant, with an implied price that has been reported at levels above $100 per megawatt-hour. Google has signed agreements with Kairos Power to buy power from a fleet of small fluoride salt-cooled reactors totaling roughly 500 megawatts. Amazon has invested in X-energy and is planning small modular reactor deployments in the Pacific Northwest. These are binding or near-binding commitments from the most sophisticated energy buyers on earth. HGP is not competing with those deals in the present tense. It is competing with them in the future tense. The Microsoft-Constellation transaction is aimed at delivery around 2028. The Google-Kairos timeline stretches into the early 2030s. The entire argument for a decommissioned naval reactor is that it can be deployed faster than a newly built small modular reactor, possibly within three to five years, because the heavy hardware already exists. That speed is HGP's only legitimate moat. Everything else is speculation. The same logic applies to the regulatory comparison. NuScale, the most mature American SMR developer, submitted its design certification application to the NRC in 2008 and did not receive approval until 2022. Fourteen years. Oklo, the Sam Altman-backed startup that also went public via SPAC, had its early design certification application rejected by the NRC in 2022 on procedural grounds. No advanced reactor company has yet completed the journey from design to commercial operation in the United States. A retrofitted naval reactor has no established licensing pathway at all. It would not start at the starting line. It would start behind it, in a parking lot where the rules have not yet been written. The ADVANCE Act of 2024 instructs the NRC to streamline licensing for advanced reactors and offers fee relief for qualified technologies. That is real and helpful. But a statutory instruction to streamline is not the same as a completed license. The NRC is an independent regulator. Its safety culture does not dissolve because Congress wants faster decisions. And no fast-track provision in the ADVANCE Act specifically contemplates the transfer of a military reactor core into commercial civilian service. That is a legal blank space, and a blank space is not a timeline. Now we come to the part that should make a crypto-native reader feel at home: the financial vehicle. HGP is reportedly approaching public markets through a SPAC. That choice is itself a piece of information. The SPAC market has a documented history with nuclear and clean energy stories. NuScale entered the public market through a SPAC. Oklo did the same. Both saw their share prices get punished when the gap between narrative and physics became impossible to ignore. Oklo is the cleanest comparable. After its SPAC debut in May 2024, the stock traded above ten dollars. By August 2024, it had fallen to roughly three dollars and fifty cents. Then the AI-plus-nuclear narrative caught fire, and by early 2025 the same stock was trading above fifty dollars. That is not an investment thesis. That is a mood ring. The volatility tells you that the market is pricing a story, not a power plant. Oklo has signed non-binding letters of intent with data center operators, including agreements that were announced with large headline numbers. Non-binding means exactly what it says. No binding commercial operation has been delivered. No NRC license has been granted. The revenue is imaginary, but the stock chart is real. NuScale offers the cautionary tale inside the cautionary tale. In 2023, its first commercial customer, a Utah municipal power consortium, cancelled the project after costs spiraled. The cancellation nearly destroyed the company's momentum. The lesson is not that small modular reactors are impossible. The lesson is that cost estimates from developers are not commitments. A decommissioned reactor may have lower initial capital cost than a new build, but it carries a fixed remaining lifetime, often only fifteen to twenty-five years of operation after conversion, plus conversion costs, licensing costs, security costs, and fuel disposition costs stacked on top. When you divide all of that by the remaining electrons in the asset's life, the economic advantage over a new reactor starts to shrink. The asset is cheap. The resurrection is not. There is an even deeper issue hiding in the SPAC structure itself. A SPAC is built on a clock. The blank-check vehicle usually has a limited window, often around twenty-four months, to complete a business combination and start deploying capital. Nuclear licensing does not respect clocks. It respects physics and precedent, and both move slowly. The mismatch between SPAC capital's patience and nuclear infrastructure's gestation period is not a design flaw in HGP's plan. It is the core risk of the entire trade. Public market investors who buy the merger story are effectively buying an option on a regulatory miracle. Now consider the channel through which this story reached my desk. The source is Crypto Briefing, a publication oriented toward blockchain and digital assets, not toward the traditional utility analyst community. That channel choice is a signal. HGP's narrative is not aimed at pension funds that have spent decades underwriting coal plants. It is aimed at a different kind of capital, the same speculative capital that financed ICOs, DeFi farms, and NFT collections. That capital is fast, impatient, and narrative-hungry. It does not read safety analysis reports. It reads headlines. The founders of this project appear to understand exactly which crowd they are feeding. I have spent more than half a decade inside that crowd. I have organized events, audited protocols, and written long-form analysis for people who believe code can reorganize power. I have watched the same dynamics repeat across asset classes. First, a structural truth appears. Then, a narrative colonizes it. Then, the narrative attracts capital that cannot tell the difference between the truth and the story. This is not a condemnation. It is a description of how technology finance works. But when the underlying asset is a nuclear reactor, the cost of narrative error is not just a worthless token. It is a poisoned regulatory environment for every serious nuclear startup that follows. The hidden stake in this story is the uranium fuel chain. Even if HGP solves every licensing problem, even if the Navy transfers the hull, even if the NRC writes a new framework, the fuel must come from somewhere. Global uranium supply is concentrated in a handful of countries: Kazakhstan over forty percent, Canada around fifteen, Namibia and Australia around ten percent each. Russia dominates conversion and enrichment capacity at levels that create systemic vulnerability. The U.S. banned Russian uranium imports in May 2024, a necessary policy that also removed a cheap source of supply. Spot uranium prices have moved from roughly thirty dollars per pound in 2020 to a range around eighty to one hundred dollars per pound in 2024. That is not a gentle trend. That is a supply chain under stress. A converted naval reactor, if it uses a fresh low-enriched core, is entering that same stressed market. If it tries to use its existing HEU core, it enters a legal and political minefield. There is no cheap path. The fuel fairy does not visit nuclear startups. She visits companies with secure supply agreements, and those agreements take years to construct. Let me now turn to the market context that is too often ignored in energy stories. We are in a bear market for speculative technology assets. The era of free capital is over. Investors are asking which projects are bleeding, not which projects are dreaming. In this environment, a $1 billion SPAC valuation for a pre-license, pre-revenue, pre-fuel nuclear project demands severe scrutiny. The only reason such a deal can be discussed at all is the AI electricity panic. That panic is genuine. Data center developers will sign almost anything to secure clean firm power. But signing a non-binding letter of intent with a SPAC-backed nuclear startup is not the same as writing a check. Hyperscalers are sophisticated. They have teams of engineers who will audit HGP's fuel pathway, licensing strategy, and remaining asset life. If those audits fail, the SPAC will not deliver power. It will deliver a shareholder lawsuit. And yet, I do not believe this story should be dismissed. The deeper signal is that the market is finally pricing what I have been calling the structural coupling between compute and clean baseload. For a decade, the energy transition narrative belonged exclusively to wind, solar, batteries, and perhaps green hydrogen at the margins. The arrival of AI load has broken that monopoly. Nuclear power, including small modular reactors and even retrofitted naval reactors, has entered the portfolio as a serious option. That is not a pro-nuclear endorsement or an anti-renewables attack. It is an engineering observation: hyperscale AI needs 24/7 carbon-free power, and the set of technologies that can provide it is small. Nuclear is in that set. The market is responding. Here is where I should note a distinction that has gotten lost in the coverage. The HGP plan is not primarily a bet on old hardware. It is a bet on institutional access. If the military-to-civilian transfer can be made to work once, it establishes a pipeline. That pipeline is worth real money because it is nearly impossible to replicate. A new entrant cannot simply order a decommissioned submarine reactor from a catalog. There is no open market. There is a government, a Navy, a classification system, and a set of relationships built over decades. HGP's true competitive advantage, if it has one, is not thermodynamic. It is relational. That is a strong moat, but it is also a fragile one. Relationships change with administrations. Policy priorities change with elections. A moat that depends on a single government relationship is only as durable as that relationship. Let me also flag the risk that this narrative becomes self-serving. In SPAC land, admitting the existence of a risk is often just a rhetorical device to move past it. The source article mentions that regulatory obstacles and economic feasibility remain key challenges. I have read enough equity research to recognize that phrase. It is the verbal equivalent of a disclaimer at the bottom of a tweet. The challenge is not mentioned to inform the reader. It is mentioned to vaccinate the narrative against criticism. The real technical questions are far more brutal. What exactly is the fuel in the core? What is the plan for spent fuel disposal? What is the security perimeter at a commercial data center site hosting a military-origin reactor? What happens to the economics if the remaining operating life is only fifteen years? What is the decommissioning fund for a decommissioned reactor that has already been recommissioned? None of these questions can be answered by a headline. During the DeFi Summer of 2020, I spent weeks modeling governance dynamics for the MakerDAO system. I watched the gap between protocol ideal and market behavior produce outcomes that no one had predicted. The people who made money were not the ones with the purest ideology. They were the ones with the most rigorous models. The same principle applies here. The people who will make rational decisions about nuclear-powered data centers will not be the ones who believe every press release. They will be the ones who model the fuel path, the license timeline, the remaining asset life, and the buyer's willingness to sign a binding contract. There is a parallel between this moment and the crypto mining industry that deserves attention. Bitcoin miners spent years hunting for stranded and cheap energy, often in places where the grid could not absorb their load. That hunt produced an entire ecosystem of energy-as-a-service companies, some of them genuine, many of them predatory. The AI data center boom is the same hunt on a larger scale, except the energy buyer is not a miner with a GPU farm. It is a trillion-dollar cloud company with a net-zero commitment. That buyer has options. It can build new gas plants and buy carbon offsets. It can overbuild wind and solar with storage. It can restart a shuttered large reactor. It can invest in an SMR developer. Or it can choose to do nothing and risk running out of compute capacity at exactly the moment when its competitor is launching the next large language model. The desperation is real. And desperation, in financial terms, is a willingness to pay a premium. The AI electricity premium is the single most important new fact in the energy economy. Historical power markets assumed that a baseload plant sells into a pool and receives a marginal market price. The new model inverts that. Data center operators are willing to sign bilateral contracts at prices far above pool rates, sometimes thirty to one hundred percent above traditional industrial tariffs, because the cost of downtime is catastrophic. That shift transforms the business model for any clean firm power asset. A nuclear plant that could not compete in an old energy-only market becomes viable in a market where the buyer is purchasing availability, not just electrons. This is why the contrarian angle in this story is not the usual one. The usual contrarian angle says HGP will never get licensed, and the valuation is absurd. That is probably true, but it is also too easy. The sharper contrarian observation is that HGP may be solving the wrong problem with the wrong asset. The hyperscaler's pain is concentration. A 100-megawatt reactor serving a 100-megawatt data center is a single point of failure wired into the heart of an AI operation. If that reactor trips, the data center falls dark. The customer needs not just clean firm power, but resilient clean firm power delivered across multiple sites. An extremely small number of retrofitted naval reactors located on one government-adjacent site cannot provide that geometric resilience. The market may be searching for a fleet, and HGP is offering a handful of bespoke conversions. The better use of the same instinct is a portfolio approach that combines new SMR designs, geothermal, advanced long-duration storage, and a small number of highly reliable reactor restarts. The Microsoft-Constellation deal already points in that direction: take an existing asset with an existing license and an existing site, and buy a twenty-year option on its output. That deal required no new regulatory framework, no HEU debate, and no hand-waving about military-to-civilian transfer. It was a commercial contract, not a science project. That is the template that will clear the market. HGP may be a pioneer, or it may be a decoy that absorbs all the regulatory risk so that later, more conventional projects can benefit. Both outcomes are valuable to the energy transition. Neither outcome makes the SPAC investor rich. I also want to address the sustainability accounting, because any honest energy analysis must measure the full life cycle. A nuclear reactor produces near-zero emissions during operation. Over its full lifecycle, nuclear power emits roughly five to twelve grams of CO2 equivalent per kilowatt-hour, comparing favorably with wind and far better than natural gas. Uranium mining and enrichment do carry an upstream footprint, and the enrichment stage is carbon-heavy if the enrichment facility buys electricity from coal plants. The spent fuel must be managed for thousands of years. None of those costs appear on a SPAC pitch slide. But they will appear on a balance sheet eventually, either as an explicit line item or as a government backstop. The carbon argument for nuclear is solid. The financial argument for retrofitted naval reactors is not yet solid. The distinction matters because the energy transition does not need false starts. It does not need another UAMPS cancellation that poisons the well for everyone. The industry needs projects that can complete, and completion requires honest timelines, honest cost estimates, and honest fuel plans. Let me offer the framework I use when analyzing any energy asset in a bear market. First, map the balance sheet. Who is paying, and what happens if they stop paying? Second, map the supply chain. Where does the fuel come from, and can a hostile government interrupt it? Third, map the regulatory clock. How many independent approvals are required, and what is the historical duration of each? Fourth, map the customer contract. Is it binding, or is it a letter of intent written by a sales team? Apply that framework to HGP, and the answers are uncomfortable. The balance sheet is a SPAC waiting to happen. The supply chain passes through a classified military inventory. The regulatory clock is undefined. The customer contract, if it exists at all, has not been disclosed. That is not an investment. That is an option on an outcome that is only partially within the company's control. But the market context deserves nuance. The article did not arrive inside a vacuum. It arrived during a period when the original source analyses have frequently been classified as new energy and carbon neutrality stories because AI data centers are now consuming so much power that they have become climate policy actors. A decommissioned reactor powering an AI cluster is not a crypto story, not purely a nuclear story, and not purely a renewables story. It is a convergence story. The convergence of compute, energy, and climate policy is going to produce many strange financial hybrids over the next decade. Some will be legitimate. Some will be shells. The shell and the egg look identical from the outside. I have spent twenty-one years observing the edges of technology finance, from quantitative modeling to protocol governance to community building. I have seen more hype cycles than I care to count, and I have seen a few genuine revolutions. The genuine ones always had the same signature: a repeatable unit that got cheaper and more reliable with deployment. Solar panels. Wind turbines. Containerized compute. The fake ones had a different signature: a scarce relic wrapped in a scarcity story. A decommissioned submarine reactor is a scarce relic. The SPAC valuation is a scarcity story. If the economics only work because the asset is scarce, the business model will never scale. If the economics work because the AI buyer values clean firm power above market rates, then the business model can scale, but it will need new reactors, not just old ones. At that point, the naval reactor is not the destination. It is the proof of concept for a much larger fleet of purpose-built small reactors. That is the only version of this story that ends well for civilization. HGP succeeds not by turning one submarine into a data center power plant, but by demonstrating that clean firm power can be deployed faster and cheaper than the incumbents believe. Even if HGP fails, the demonstration value may persist. The idea of converting existing nuclear assets, restarting shuttered plants, co-locating reactors with gigantic loads, and writing long-term contracts at premium prices, that idea has already escaped into the market. It is moving through boardrooms in Redmond, Mountain View, and Seattle. It will not be stopped by a single SPAC failure. And here is the note I want every reader to hold. The HGP story is being told in a bear market, during a moment when investors are afraid of losing principal, not missing upside. In such a market, survival matters more than gains. This applies to projects as well as portfolios. The projects that survive will be the ones with binding customer commitments, transparent regulatory strategy, and fuel supply that does not depend on an act of Congress. HGP has none of those things publicly demonstrated. Perhaps the private disclosures are stronger. Perhaps the Navy conversations are further along than published reports suggest. Perhaps the SPAC will find a buyer who has deep enough pockets to wait a decade for licensing. That is possible. But possibility is not probability, and a $1 billion valuation should reflect probability, not possibility. Trust no one. Verify everything. I do not mean that as a call to cynicism. I mean it as a call to rigor. The energy transition is too important to be left to narrative alone. If you are evaluating this project, do not ask whether submarine reactors are cool. Ask whether the fuel path is legal. Ask whether the license timeline fits the SPAC clock. Ask whether the data center buyer will accept a single point of failure. Ask whether the remaining asset life can amortize the conversion cost. Those are the questions that separate investors from gamblers. There is an old saying in the nuclear community that is worth repeating here. Gold is heavy. Code is light. A blockchain application can be deployed in an afternoon and abandoned by midnight. A nuclear reactor cannot. It demands respect, time, and institutional memory. The attempt to graft silicon-valley speed onto naval-reactor physics is the most interesting experiment of this cycle, and the most dangerous. If it works, it could unlock a new era of clean firm power for AI. If it fails, it will hand a free gift to every skeptic who says distributed nuclear is a fantasy. I hope the engineers get more time than the marketers. I hope the NRC gets more budget than the merger lawyers. And I hope the next time someone sends me a Crypto Briefing story about a $1 billion nuclear SPAC, it comes with a binding power purchase agreement attached. Summer fades. Builders remain. The same is true of reactors. The hull will rust. The license will expire. The data center will eventually become a ruin. What remains is the underlying architecture, the electric bones of a world that chose to power its intelligence with clean atoms and clean electrons. Whether HGP is part of that architecture or a footnote to it will depend not on the press release, but on the paperwork that follows. Watch the filings, not the headlines. Watch the fuel contracts, not the keynote speeches. Watch for a single binding letter from a hyperscaler with a real token and a real penalty clause. Until that letter exists, this story is a signal, not a stock. And in the silence after the SPAC announcement, the loudest signal in the room is the absence of that letter. Noise is cheap. Signal is rare. This is still mostly noise. But underneath it, buried like a reactor in a steel hull, is a signal about the future of energy that no amount of regulatory failure can erase. The grid was built for a world of predictable demand. It is now being asked to feed a machine that never sleeps. Every technology that can answer that demand without burning the climate is about to be repriced. Decommissioned reactors, new SMRs, geothermal wells, and long-duration batteries are all now competing for the same premium dollar. The premium exists because the demand is real. The winners will be those who can convert real demand into deliverable power before the market's patience expires. HGP may not win. But the race it is entering is the most important race of the next two decades. I will be watching from Berlin, calculator in hand, skeptical as always, hopeful despite myself. Because in the end, the question is not whether venture capital can afford to wait for nuclear. The question is whether the climate and the compute economy can afford anything else. The answer to that question is still being written, one license application at a time. Trust no one. Verify everything. Then go build something that deserves verification.