Tag: AI data centers

  • The Geysers: Inside the World’s Largest Geothermal Complex in 2026

    On January 7, 2026, Constellation Energy closed its $16.4 billion acquisition of Calpine Corporation, creating the largest private-sector power producer in the United States — 55 gigawatts of combined capacity, the country’s biggest nuclear fleet stitched together with Calpine’s natural-gas turbines and one unusual asset that the press release referred to as the “crown jewel of geothermal energy.” That asset is a 45-square-mile patch of the Mayacamas Mountains north of San Francisco where 18 power plants sit on top of a 1.3-million-year-old blob of cooling magma, drilling holes up to 12,900 feet deep into a sandstone reservoir to capture the steam that rises through it. The complex is called The Geysers, which is a misnomer — there are no actual geysers, never have been — and it is the largest developed geothermal field on Earth. It has been generating commercial electricity since September 25, 1960, which makes it older than the Beatles’ first single. It currently produces around 725 megawatts of around-the-clock baseload power, which is enough to run a city the size of San Francisco, which is the city it largely runs.

    It is also, in 2026, suddenly the most interesting piece of legacy infrastructure in the American energy system — because the AI build-out is consuming power faster than the grid knows how to supply it, every additional megawatt of AI computation requires roughly 1,000 metric tons of copper to deliver, the chips inside those data centers are running flat-out 24 hours a day, and “around-the-clock carbon-free firm baseload” is the rarest combination of words in the electricity business. The Geysers does it. Has done it. For 65 years. In Sonoma County. While almost nobody noticed.

    What it actually is

    The Geysers sits in Sonoma and Lake counties at the northern end of the Mayacamas range, about 75 miles north of San Francisco. Beneath it, a body of silica-rich magma intruded into the crust roughly 1.3 million years ago and never fully cooled. The rock above the magma — fractured sandstone and metamorphic graywacke — is hot enough to flash-boil any groundwater that reaches it, and the reservoir produces something extraordinarily rare in the geothermal world: dry steam. Most geothermal resources produce a wet mixture of brine and vapor that has to be separated, condensed, and processed before it can drive a turbine. The Geysers produces vapor-dominated steam at roughly 240 degrees Celsius that comes out of the wellhead ready to push a blade. The metallurgy that allows turbine blades to spin year after year in continuous high-temperature steam without failing is genuinely difficult engineering — not as exotic as the single-crystal nickel superalloys in a jet engine, but the same family of material science applied to a different operating envelope. There are only two large dry-steam fields known on the planet. The other one is at Larderello in Tuscany, which started commercial geothermal production in 1913 and inspired the entire global industry. Larderello is roughly a third of the size of The Geysers in terms of installed capacity.

    The site covers 29,000 acres of mountain ridges threaded with pipelines, well pads, generating stations, and cooling towers venting white steam plumes that are visible from passing aircraft. There are 591 wells in total, 376 of them currently active, drilled in some cases more than two miles below the surface. The steam is piped — through insulated, above-ground steel lines that snake across the ridgelines for miles, an exoskeleton of metal threading through chaparral the way the steel pipes of the Paris pneumatic post network once threaded through a city — to a network of 18 generating facilities, where it spins turbines, gets condensed into liquid water in cooling towers, and is then re-injected back into the reservoir to keep the cycle running. The whole system runs at a capacity factor in the low-to-mid 50s, meaning it is generating actual electricity roughly 53 percent of the time it could theoretically be running — extraordinary uptime for a renewable resource, dramatically better than solar’s 25 percent or onshore wind’s 35 percent, and the entire reason every hyperscaler in California is suddenly interested in geothermal power.

    The decline and the toilet-flush rescue

    The Geysers nearly killed itself in the 1980s. The original 1960 plant — PG&E Unit 1, a modest 11-megawatt machine on Big Sulphur Creek — proved the concept, and over the next three decades the industry expanded aggressively, adding plant after plant, drilling well after well, pulling steam out of the reservoir faster than the slow trickle of rainfall could replenish it. Total installed nameplate capacity climbed past 2,000 megawatts. Then, by 1989, the steam pressure began to drop. Wells that had once roared started to wheeze. Production declined. Several plants ran at a fraction of nameplate. Calpine shut down Units 9 and 10 entirely in 2000 and 2001 because they couldn’t make money on the steam they were getting. The reservoir, it turned out, was a finite bathtub being drained faster than the spigot filled it, and the geothermal industry’s marquee American project was on a glide path to extinction. The Romans built aqueducts on the same assumption and ran them dry. The Mayan farmers cleared rainforest on the same assumption and watched their soil collapse. The Geysers’ operators built power plants on the assumption that the rain would keep up with the wells, and the rain did not.

    The fix, when it came, was so unlikely that nobody outside the geothermal industry seems to know about it. In 1997, a consortium of operators — Calpine, NCPA, and Unocal at the time — completed the Southeast Geysers Effluent Pipeline (SEGEP), a 29-mile (later 40-mile) pipeline that carries treated wastewater from Lake County sewage plants up the mountain and pumps it down injection wells at a rate of about 9 million gallons per day. The water hits the hot reservoir, flashes to steam, rises through the production wells, drives the turbines, and gets condensed back into water in the cooling towers. The system was supplemented in 2003 by the Santa Rosa Geysers Recharge Project (SRGRP), a $250 million, 41-mile pipeline that climbs 3,000 feet through residential developments, vineyards, and the Mayacamas Mountain Sanctuary owned by the Audubon Society — which sued, then settled for $1.3 million — to deliver another 11 million gallons per day of tertiary-treated wastewater from Santa Rosa, Rohnert Park, Cotati, and Sebastopol. Total injection now runs around 20 million gallons of recycled water per day, which absorbs roughly 65 percent of the treated effluent from those communities and supports an estimated 77 megawatts of generation capacity that would not otherwise exist. Sonoma County’s flushed toilets are, in a quite literal sense, powering the data centers in San Francisco.

    The SRGRP project worked exactly as designed. Steam pressure stabilized. Production declines slowed. The reservoir, which had been on a glide path to commercial death, is now expected to continue producing into the 2070s and possibly beyond. The closed-loop engineering — a system that runs on its own waste streams and produces its own inputs — is the same logic the ancient Persian qanats achieved through gravity 2,500 years ago and that Barcelona’s pneumatic waste network now achieves by piping garbage to a power plant at 70 kilometers per hour. The difference here is that the input is sewage and the output is gigawatt-hours, which is a thermodynamic transformation so unromantic that none of the parties involved is interested in publicizing it. A federal judge, two sanitation districts, and a geothermal company built the modern version of a closed-loop irrigation system on a hillside in California, and the customers paying premium rates for around-the-clock renewable electricity have, by and large, no idea.

    The earthquakes the city tries not to mention

    Injecting cold water into hot fractured rock makes the rock crack. The cracks make small earthquakes. The Geysers has, since the SEGEP and SRGRP pipelines came online, generated roughly 4,000 microearthquakes per year in the magnitude 1-to-3 range, with occasional events climbing into the magnitude 4 range that residents within a 20-mile radius can absolutely feel. The largest event attributed to Geysers injection was a magnitude 4.6 in 2006. None have caused significant structural damage. Almost all are tied directly, in the seismic record, to the volume and rate of wastewater injection in specific zones of the field. The operators monitor the swarms in real time and have learned over two decades to throttle injection at specific wells when activity climbs above thresholds.

    The locals are not uniformly thrilled. “It’s Santa Rosa’s wastewater, and they don’t feel the earthquakes,” one Lake County resident told Scientific American in the early years of the SRGRP. The trade-off is asymmetric — the costs are borne by 500 year-round residents of the immediate area, the benefits flow to 725,000 households spread across five counties and increasingly to the AI training runs being conducted in server farms hundreds of miles away. It is the same distribution pattern that gives the Mumbai dabbawalas their famous reliability — costs concentrated in one place, benefits distributed across another — except that nobody is grumbling about the dabbawalas because their externalities are confined to the labor market they operate inside. The Geysers exports its electricity and imposes its tremors, the two flows do not move through the same zip codes, and the asymmetry is what makes mineral-rich communities carry the costs of energy systems consumed by people they will never meet.

    The Constellation acquisition and the AI dimension

    The reason any of this matters in 2026 in a way it did not matter in 2024 is that the calculus of electricity has changed. The Constellation-Calpine merger, which closed January 7, 2026, with a $16.4 billion equity price and a $26.6 billion enterprise value once Calpine’s debt was rolled in, was not really about geothermal — it was about Constellation building what its CEO Joe Dominguez has called a “one-stop shop” for the AI data center boom. The combined company has 32.4 gigawatts of nuclear, 26 gigawatts of natural gas, and the Geysers’ ~725 megawatts of geothermal — enough firm, around-the-clock generation to underwrite the kind of multi-year power purchase agreements that Microsoft, Google, Meta, and Amazon are now signing with anyone who can credibly promise to keep the lights on for a hyperscale training cluster. The nuclear side of the merged portfolio is the bigger story in absolute megawatts — Constellation is the country’s largest reactor operator at a moment when the uranium supply chain is scrambling to fuel a nuclear renaissance it was not prepared for — but the geothermal side is the more interesting one structurally, because it sits at the intersection of a 65-year-old operating asset and a brand-new demand vector. Constellation’s stock has nearly doubled since the deal was first announced in January 2025. Analysts are forecasting a 20 percent boost to 2026 earnings per share. The Department of Justice forced the divestiture of two Texas natural-gas plants — the Jack A. Fusco Energy Center near Houston and the Gregory Power Plant near Corpus Christi — to clear antitrust review, the first DOJ consent decree in a major U.S. electricity merger in fourteen years.

    What this means for The Geysers is that, for the first time in its commercial history, the site is part of a company whose primary strategic question is “how do we sell more electrons to people running language models.” The answer for The Geysers itself is “we make more steam,” and that is exactly what is happening. Calpine’s North Geysers Incremental Development (NGID) project, which began phased completion in 2025 and is scheduled for full commissioning in June 2026, is drilling 11 new production wells and 2 new injection wells across four well pads in the northern portion of the field, using existing pipeline infrastructure to route the additional steam to nearby plants. The first 7 megawatts of incremental output came online in June 2025 and is being purchased by MCE, the Bay Area community-choice aggregator. The full 25-megawatt expansion will be online by mid-2026 — modest in absolute terms compared to a 1.5-gigawatt natural-gas plant, but each of those 25 megawatts produces emissions-free electricity at a capacity factor that no solar farm or wind project can touch, which is exactly what the data center buyers want.

    The longer-term story, though, is that The Geysers is no longer the only geothermal game in town. Enhanced Geothermal Systems — the new class of next-generation drilling technology pioneered by companies like Fervo Energy and Sage Geosystems — uses horizontal drilling and hydraulic stimulation techniques borrowed wholesale from the fracking industry to create artificial geothermal reservoirs in hot dry rock formations that don’t have naturally occurring hydrothermal systems. The underlying technology is the same directional-drilling apparatus that revolutionized shale oil and that runs on a global supply chain of specialty metals and electronic components that overlaps almost completely with the supply chain feeding the energy-intensive industries on the other side of the meter. Fervo’s Cape Station project in Beaver County, Utah is scheduled to deliver its first 100 megawatts of commercial output in 2026 and ramp to 500 megawatts by 2028. The company went public on May 13, 2026 at $27 per share and closed its first trading day up 33 percent, raising $1.89 billion to fund construction. Google, Meta, and a half-dozen other hyperscalers have signed offtake agreements for the output. Fervo’s pitch — and it is a credible one — is that EGS can deliver geothermal power outside the small handful of geological accidents like The Geysers, which means a future in which geothermal might supply not 0.4 percent of U.S. electricity but a meaningful fraction of total demand. One projection from Project InnerSpace estimates that geothermal could cover 64 percent of AI data center energy demand by 2030 under aggressive deployment scenarios.

    Geothermal is competing for the same data center power purchase agreements that the nuclear fusion industry is now chasing — Commonwealth Fusion’s SPARC, Helion’s commercial plant, TAE’s commercial program — and that conventional fusion still hasn’t delivered after 70 years of being thirty years away. The pitch for each technology is roughly identical: around-the-clock carbon-free firm baseload power at gigawatt scale. Geothermal’s advantage is that it works today, at this site, with this engineering, and has been working since 1960. Its disadvantage is that the resource is geographically constrained to a small number of places on Earth, and the new generation of EGS deployments has not yet proven that the constraint can be engineered around at scale. The fusion companies have the opposite problem — the resource is universal but the technology has not yet produced a single commercial kilowatt-hour.

    What 2026 actually looks like up there

    If you stand on a ridge in the Mayacamas in 2026 and look down at the steam plumes, what you are seeing is an industrial complex that has been continuously operating since the year John F. Kennedy was elected president, that has survived its own near-death from over-pumping, that recovered through a 41-mile sewage pipeline running uphill through wine country, that generates 4,000 small earthquakes a year as a side effect of that recovery, that supports roughly 300 Calpine employees and 150 contractors who live in Lake and Sonoma counties, and that has just been folded into the corporate balance sheet of the largest private power producer in the United States in a $26.6 billion transaction whose primary justification was the energy needs of artificial intelligence. The 18 plants stay in operation while individual wells are deepened, while injection rates are throttled to manage seismicity, while new pipelines are commissioned, while the corporate ownership above them changes — the same maintain-while-running discipline that keeps an 800-meter outdoor escalator moving through Hong Kong typhoon seasons or that keeps Manhattan’s 144-year-old steam grid heating skyscrapers whose architecture made conversion to other fuels economically impossible.

    The whole system is, in the most literal infrastructure sense, a survivor — the kind of installation that was built in a moment of technological enthusiasm, declined into commercial near-obsolescence, retrofitted itself with an unlikely fix that almost nobody would have predicted, and now sits at the center of a strategic conversation about whether the growth in computing demand and the growth in electrified manufacturing can be supplied by an electricity sector that was, until very recently, planning for flat demand. The same survival logic applies to pneumatic networks under Berlin that have outlasted five regimes, to the single rotating boat lift in Scotland that revived a derelict canal network, and to the suspended monorail in Wuppertal that has been carrying commuters over a German valley since 1901. Infrastructure does not have to be efficient or fashionable or new to be valuable. It has to be there. The Geysers has been there for 65 years, in dry steam, beneath 591 wells, on top of a 1.3-million-year-old magma intrusion, powering a city full of people who do not know its name — and it has never been more economically valuable than it is right now, in the middle of 2026, with twenty-five new megawatts of capacity coming online, a new owner whose entire corporate strategy is built around selling firm power to hyperscalers, and a generation of competing geothermal technologies trying to do, somewhere else in the western United States, what the Mayacamas Mountains have been doing on their own, since 1960, for free.

  • The Copper Shortage in 2026: Why the Energy Transition Can’t Work Without It

    Copper hit $13,240 per metric ton on the London Metal Exchange in January 2026 — a record. The price had risen nearly 40 percent in 2025 alone, its largest annual gain since 2009. And the deficit hasn’t started yet. BloombergNEF projects that the copper market enters structural deficit in 2026, meaning global demand permanently exceeds the ability of mines to supply it. S&P Global’s January 2026 study, “Copper in the Age of AI,” projects demand will reach 42 million metric tons by 2040 — a 50 percent increase from current levels — while production peaks at 33 million metric tons in 2030 and then declines. The resulting shortfall: 10 million metric tons by 2040, roughly 25 percent below projected demand. J.P. Morgan forecasts a refined copper deficit of approximately 330,000 metric tons in 2026, pushing prices potentially above $12,000 per metric ton. The market for the metal that makes electrification physically possible is about to run out of the metal.

    Why copper is different from every other critical mineral

    Copper isn’t rare. It’s the third most-used industrial metal on earth after iron and aluminum. It exists in economically extractable concentrations on every continent. There is no geographic monopoly — Chile, Peru, the DRC, China, the United States, and Australia all produce significant quantities. The copper shortage in 2026 is not a concentration problem the way gallium (98 percent China) or rare earth processing (90 percent China) are concentration problems. It’s a volume problem. The world needs more copper than it can produce, and the gap between the two is widening.

    An electric vehicle uses 80 to 100 kilograms of copper — three to four times what a conventional car uses — concentrated in the motor, battery, power electronics, and charging system. A single large offshore wind turbine contains roughly 8 metric tons of copper in its generator, transformer, cabling, and grid connection. A Level 3 fast-charging station requires substantial copper for high-voltage connections and power conditioning. Solar installations, grid-scale battery storage, power distribution networks, and the transformer substations that connect renewable generation to the grid all run on copper. An AI data center requires over 1,000 metric tons of copper per facility. Grid expansion alone — the wiring that connects everything — accounts for the largest single category of copper demand growth through 2050.

    Daniel Yergin, vice chairman of S&P Global, summarized the problem in the study’s opening: copper is the great enabler of electrification, but the accelerating pace of electrification is an increasing challenge for copper. EVs, grid expansion, renewables, AI data centers, digital infrastructure, and defense spending are all scaling simultaneously. Supply is not on track to keep pace. The question is whether copper remains an enabler of progress or becomes a bottleneck.

    Why supply can’t respond

    The average timeline from copper discovery to production is 17 years. In the United States, it averages close to 29 years. Chile has 13 new copper projects valued at $14.8 billion in the pipeline — most won’t produce meaningful output until 2028 or 2029. Opening a copper mine in a developed country requires exploration, feasibility studies, environmental impact assessment, permitting, judicial review (often multiple rounds), construction, commissioning, and ramp-up. Each step takes years. Environmental opposition and community resistance add additional years.

    Ore grades are declining. The average copper ore grade has fallen from roughly 1.5 percent in the 1990s to below 0.6 percent today, meaning miners move more than twice as much rock per ton of copper produced. Rising energy costs, labor costs, and water scarcity in major mining regions (Chile’s Atacama, Peru’s highlands) compound the cost escalation. Indonesia’s Grasberg mine — one of the world’s largest — is undergoing the transition from open pit to underground block caving, which temporarily reduces output during the transition. Indonesian export policy changes and domestic processing requirements further constrain material available for international markets.

    Mining companies are responding by extending existing mines rather than developing new ones. Capital for exploration and new mine development peaked at $26 billion in 2013 and roughly halved since then. BHP, Anglo American, Rio Tinto, Glencore, and Zijin have shifted capital expenditure toward copper — BHP’s copper revenue share rose from 27 percent to 38 percent between 2020 and 2024 — but the spending is going into optimizing existing operations, not building greenfield mines. The M&A activity is enormous: Glencore committed $16 billion to projects in Argentina, BHP’s attempted acquisition of Anglo American was motivated primarily by copper exposure. But buying existing mines doesn’t create new supply. It consolidates control over supply that already exists.

    Recycling helps but doesn’t close the gap

    Recycled copper currently contributes roughly 4 million metric tons annually — about 16 percent of total supply. S&P Global projects recycling will more than double to 10 million metric tons by 2040. That’s genuine progress. But the doubling of recycled supply is already factored into the 10-million-ton shortfall projection. Without the recycling increase, the deficit would be 16 million metric tons, not 10. Recycling is a structural supplement. It isn’t a substitute for mining, and it can’t close a gap measured in millions of metric tons per year.

    The copper in an electric vehicle motor won’t be available for recycling for 12 to 15 years. The copper in grid infrastructure has a lifespan measured in decades. The copper in buildings lasts longer than the buildings. The feedstock problem is the same one that constrains rare earth recycling: the products containing the material haven’t reached end of life yet, so the recyclable supply won’t arrive for years.

    The AI demand nobody modeled

    The demand driver that makes the copper shortage in 2026 categorically different from previous copper deficits is artificial intelligence. A single large AI data center requires over 1,000 metric tons of copper — power cabling, cooling systems, server racks, transformer connections, UPS systems, grid integration. Microsoft, Google, Amazon, and Meta are collectively building hundreds of these facilities. The electricity demand from AI computation is projected to grow faster than any other category of electricity consumption through 2040, and every megawatt of AI power consumption requires copper to deliver, condition, and distribute.

    The S&P Global study explicitly identifies AI as a new demand vector that previous copper forecasts did not account for. Defense spending is another: guided weapons systems, electronic warfare equipment, naval vessels, and military communications infrastructure all have rising copper intensity. Grid expansion to support both AI data centers and electrified transport is the multiplier — the infrastructure that connects new demand to new generation capacity is itself copper-intensive.

    The price signal problem

    Copper prices above $12,000 per metric ton should, in theory, incentivize new mine development. They do — eventually. But the response time is measured in decades, not quarters. A mine that receives approval today won’t produce copper until the 2030s. The price signal is operating on a timeline that is structurally mismatched with the investment cycle. Miners want sustained high prices before committing multi-billion-dollar capital. Investors want certainty that demand projections will hold. The projects themselves take 15 to 29 years to develop. The deficit builds during the interval.

    There is also a narrative problem. BloombergNEF’s Kwasi Ampofo calls the copper shortage structural, not cyclical. But some analysts push back: copper mining companies have been effective at promoting a long-term shortage narrative, and markets may have priced in future scarcity prematurely. Nearly one million metric tons of copper are reportedly parked in U.S. warehouses, partially driven by tariff hedging rather than genuine physical tightness. The 2025 price surge was driven as much by the “EV-AI-energy transition” investment narrative as by immediate supply scarcity. Both the shortage forecast and the concern that the forecast is self-serving exist simultaneously, which is the kind of epistemic situation the critical minerals space generates constantly.

    What it means

    Six countries produce roughly two-thirds of mined copper. The supply chain isn’t as concentrated as gallium or rare earths, but it’s concentrated enough that disruptions in Chile (strikes, water policy), Peru (political instability), Indonesia (export rules, mine transitions), or the DRC (conflict, as the cobalt post documented) cascade through global markets. The U.S. designated copper a critical mineral in 2025. The Inflation Reduction Act directed over $30 billion toward critical mineral supply chains. None of this changes the fundamental constraint: opening new mines takes longer than the demand growth projections allow.

    The copper shortage in 2026 is the clearest case of a material where the energy transition creates the demand that the energy transition depends on, and the supply chain that served a 28-million-ton-per-year world is not structured to serve a 42-million-ton-per-year world. The gap between those two numbers is where the transition either succeeds or stalls.

    We cover the copper shortage alongside gallium export controls, the helium crisis, and the full landscape of critical materials that modern technology depends on across our Rare Earth Elements course — including why the most abundant critical metal on earth is the one most likely to constrain everything else.

  • Uranium Supply Chain 2026: Nuclear Renaissance Meets Mining Reality

    The United States operates 93 nuclear reactors — the largest fleet on earth — and cannot fuel a single one with domestically sourced uranium. The country has essentially no primary uranium production. The mines that once operated in Wyoming, Texas, and the Colorado Plateau shut down decades ago when prices collapsed, and the supply chain that supported them — the skilled labor, the processing infrastructure, the regulatory pipelines — dissolved with them. In 2026, spot uranium is approaching $92 per pound. Analysts project prices reaching $100 to $120 per pound, with some upside scenarios targeting $135 if supply fails to respond. The U.S. government has committed up to $80 billion to build new reactors and reinvigorate the nuclear industrial base. The USGS added uranium to its 2025 Critical Minerals List for the first time in years. The IEA forecasts annual nuclear investment rising from over $70 billion today to approximately $210 billion by the mid-2030s. Roughly 65 reactors are under construction worldwide.

    The demand story is real. The supply story is the problem.

    Where the uranium comes from

    Global reactor demand runs approximately 67,500 metric tons of uranium per year. Mine production has historically met only 74 to 90 percent of that, with the deficit covered by drawdowns from government and commercial inventories, recycled material, and secondary supply. Those secondary sources are depleting. The market is transitioning from an inventory-driven system to a production-driven one, and production isn’t keeping up.

    Kazakhstan dominates. Kazatomprom, the state-owned producer, is the world’s largest uranium miner, operating primarily through in-situ recovery — a technique that pumps acidified solution into uranium-bearing rock formations underground and extracts the dissolved uranium without conventional mining. Kazakhstan accounts for roughly 40 percent of global production. Canada’s Cameco operates McArthur River and Cigar Lake in Saskatchewan’s Athabasca Basin — two of the highest-grade uranium deposits on earth, with licensed capacity of 25 million pounds annually and proven reserves exceeding 457 million pounds. Australia, Namibia, Uzbekistan, and Niger round out the major producers. Russia controls a significant share of global uranium enrichment and conversion — the processing steps between mining raw uranium and fabricating reactor fuel.

    The concentration is the vulnerability. A large proportion of uranium production sits in non-Western jurisdictions. Sanctions, export bans, and the war in Ukraine are constraining the nuclear fuel cycle. Niger — historically a significant supplier — produced no uranium at all in 2025 after a military junta seized power and disrupted operations at the SOMAÏR facility. Kazakhstan has announced lower production targets for 2026. McArthur River reduced its 2025 output due to development delays. In the United States, several in-situ recovery restarts have ramped up more slowly than planned. The net effect: tighter global supply for reliable primary production, at precisely the moment when demand forecasts keep getting revised upward.

    The demand surge

    Three forces are converging on uranium demand simultaneously.

    The first is reactor life extensions and restarts. Existing nuclear plants that were scheduled for retirement are getting new operating licenses instead. Plants that were shut down are being evaluated for restart. The economics shifted when natural gas prices spiked, renewable intermittency proved harder to manage than projected, and carbon-free baseload generation became a policy priority rather than a political liability. Nuclear went from a technology that governments were phasing out to one they’re subsidizing.

    The second is new reactor construction. The $80 billion U.S. government commitment includes Westinghouse AP1000 deployments and GE Hitachi BWRX-300 small modular reactors. Canada has broken ground on SMRs at the Darlington nuclear station with combined funding commitments of roughly CAD 3 billion and a target completion around 2030. The U.S. and Japan announced a framework totaling $550 billion, with up to $332 billion directed to energy and AI-linked infrastructure including new nuclear capacity. China continues building reactors at a pace no other country matches and purchasing uranium in large quantities to stockpile for its future fleet.

    The third is AI data centers. This is the demand driver that didn’t exist in anyone’s forecast five years ago. Hyperscale computing facilities require baseload power — reliable, 24/7 generation that doesn’t depend on weather or time of day. Nuclear fits that requirement better than any other carbon-free source. More than 63 percent of investors surveyed by Uranium.io believe AI-related electricity consumption will become a material factor in nuclear planning over the next decade. Microsoft, Amazon, and Google have all explored or announced nuclear power agreements for data center operations. The AI demand signal is being treated as structural rather than cyclical — permanent new load on the grid that requires permanent new generation.

    Why supply can’t respond quickly

    This is the constraint that the nuclear renaissance runs into. Uranium mining is not a faucet. Mine restarts require years, not months. The lead time from decision to production involves permitting (often multi-year regulatory processes), environmental review, workforce recruitment (specialized uranium mining labor that largely doesn’t exist anymore in the West), facility construction or refurbishment, and ramp-up testing. A mine that was shuttered in 2012 can’t resume production in 2026 just because the price is right.

    Beyond mining, the fuel cycle has its own bottlenecks. Mined uranium (yellowcake) must be converted to uranium hexafluoride, enriched to increase the concentration of fissile U-235, fabricated into fuel assemblies, and delivered to the reactor. Russia controls a significant share of global enrichment and conversion capacity. The U.S. ban on Russian uranium imports — signed into law in 2024 — created a scramble for alternative enrichment services. Centrus Energy is the only licensed producer of High-Assay Low-Enriched Uranium (HALEU) in the Western world — the next-generation fuel that advanced reactors and many SMR designs require. Centrus is expanding its Piketon, Ohio facility, but scaling enrichment infrastructure is measured in years and billions of dollars, not quarters.

    The structural reality: even sustained high prices may not resolve supply deficits within typical investment horizons. Producers have signaled that three-digit prices per pound — above $100 — are the minimum necessary to incentivize new mine development at a scale that reflects actual capital costs, permitting timelines, and supply chain risk. The market is in a standoff. Utilities want to buy at current prices. Producers want higher prices before committing capital to new production. China, meanwhile, continues buying at whatever price the market offers, building strategic reserves while Western utilities defer purchases and hope prices stabilize.

    The SMR fuel problem

    Small modular reactors are the technology that’s supposed to make nuclear faster, cheaper, and more deployable. The first SMRs won’t be operational until 2030 or 2031. The World Nuclear Association projects SMR capacity could account for roughly 7 percent of global nuclear power generation by 2040. But many advanced SMR designs require HALEU — uranium enriched to between 5 and 20 percent U-235, compared to the 3 to 5 percent used in conventional reactors. HALEU production capacity in the Western world is essentially nonexistent outside of Centrus’s pilot-scale operations. Russia was the primary commercial supplier of HALEU before sanctions disrupted the trade.

    Building the SMR fleet without the fuel to power it is the kind of sequencing error that turns a technology roadmap into a bottleneck cascade. The reactors require enrichment capacity that requires enrichment facilities that require regulatory approval that requires years. Cameco’s $2.8 billion ten-year supply agreement with India and Centrus’s $1.2 billion in convertible note offerings and $2 billion in contingent utility purchase commitments represent the financial architecture being constructed to close these gaps. Whether the construction finishes before the demand arrives is the open question.

    The investment case and the honesty test

    Uranium is one of the few commodities where there is essentially no substitution potential. A nuclear reactor runs on uranium. Nothing else does the job. Demand is inelastic — utilities will pay whatever the market requires because the cost of uranium is roughly 5 to 7 percent of a reactor’s total operating budget. A doubling of uranium prices is a rounding error in the cost of nuclear electricity. This means utilities will eventually buy at higher prices because they have no alternative. The question is when, not whether.

    Long-term contract prices have risen to $86 per pound, indicating that utilities are accepting elevated costs even as they resist spot purchases. The World Nuclear Association has revised its uranium demand growth forecast to a 5.3 percent compound annual growth rate through 2040, up from 4.1 percent previously. Analysts project a supply deficit building over the next decade as mine production continues to lag reactor requirements. More than 85 percent of surveyed investors anticipate higher prices into 2026.

    The honesty test: every part of this demand story — reactor restarts, new construction, SMRs, AI data centers — requires uranium, and the supply chain to deliver it doesn’t exist at the scale the demand forecasts imply. The nuclear fuel supply chain is being rebuilt in real time, by governments writing checks and producers scaling operations, against a backdrop of geopolitical disruption, depleted inventories, and a workforce that needs to be reconstituted essentially from scratch in the West. The nuclear renaissance is real. The mining and enrichment infrastructure to fuel it is years behind.

    We cover the uranium supply chain alongside gallium and germanium export controls, the helium shortage, and the full landscape of critical materials that modern technology and energy systems depend on across our Rare Earth Elements course — including why the largest nuclear fleet on earth can’t fuel itself, and what that means for a planet betting on reactors to keep the lights on.