Uranium Mining and the
Nuclear Energy Revival

The supply-demand balance that every bank, every consultancy, every fund publishes depends on an estimate of how much uranium is sitting in warehouses, government vaults, and utility fuel stores around the world. That estimate is, across the board, unreliable.

Start with China. The strategic uranium reserve maintained by CNNC and its affiliates has never been publicly quantified. The range circulating in the industry, something like 30,000 to 100,000 tonnes, is sourced from inference and secondhand reporting, not from any official disclosure. UxC and TradeTech, the two specialized consultancies whose data underpins most commercial uranium analysis, both acknowledge the uncertainty in their methodology notes but still have to put a point estimate into their models. The gap between the low and high end of the China estimate alone is larger than total global mine output for a full year.

Japan is a different kind of opacity. Japanese utilities held substantial fuel inventories when the fleet shut down post-Fukushima. Some of that material was consumed by the handful of reactors that restarted. Some was committed under existing contracts and may have been resold or swapped. Some may still be sitting in storage at conversion and fabrication facilities in Europe and North America under custodial arrangements. The Euratom Supply Agency's annual report provides aggregate European inventory data but it is backward-looking and does not break out Japanese-owned material stored in EU facilities. The U.S. EIA's Uranium Marketing Annual Report covers domestic utility inventories but has limited visibility into material held by non-utility entities, traders, and financial funds.

The WNA's Nuclear Fuel Report, published every two years, is the closest thing the industry has to a comprehensive supply-demand reference. The 2023 edition projected a growing gap between reactor requirements and primary mine supply through the 2030s. But the report's treatment of secondary supply and inventory drawdown relies on assumptions that the WNA itself flags as uncertain. When the central scenario of the industry's most authoritative reference document contains a multi-thousand-tonne uncertainty band driven by inventory unknowns, the confidence intervals on any price forecast derived from it should be much wider than what typically gets presented to investors.

If the true global inventory is at the low end of estimates, the market is already critically tight. If it is at the high end, there is a cushion that could absorb several years of the mine-production-to-reactor-demand deficit before prices face sustained structural pressure. The difference between those two scenarios is the difference between uranium at $80 and uranium at $200. And the data to distinguish them does not exist in the public domain.

The grid math case for nuclear has been sitting there for fifteen years. Capacity factors above 90% versus 25%-35% for wind, 15%-25% for solar. Firm, dispatchable, zero-carbon generation at gigawatt scale. None of this is new information.

What broke through the political resistance was the convergence of three things that all landed in the same two-year window. The European energy crisis of 2022 destroyed confidence in the gas-as-bridge model. AI load growth gave American politicians a constituency (tech companies) that wanted nuclear and had lobbying budgets to match. And climate targets set in 2015-2016 started hitting their intermediate checkpoints, making the gap between emissions pledges and actual generation mix impossible to wave away.

None of which changes the fact that nuclear plants are notoriously difficult to build on time and on budget in Western countries. Vogtle Units 3 and 4 in Georgia came in at roughly double the original cost estimate and years late. Olkiluoto 3 in Finland took eighteen years from construction start to commercial operation. Flamanville 3 in France is a similar story. Hinkley Point C in the UK is tracking over budget and behind schedule. The contrast with China, where CGN and CNNC routinely deliver units on schedule and near budget, is stark and uncomfortable for anyone arguing that nuclear construction is inherently unmanageable.

The part of the uranium story where the largest analytical errors get made. Almost all mainstream coverage treats "uranium price" as the relevant variable. The relevant variables are conversion price, enrichment price, and the availability of enrichment capacity denominated in SWU, particularly non-Russian SWU.

The sequence: mine produces yellowcake (U₃O₈). Yellowcake goes to a conversion plant and becomes UF₆. UF₆ goes to an enrichment plant where centrifuges raise the U-235 fraction from 0.711% to 3%-5%. Enriched UF₆ goes to a fuel fabricator. Fuel assemblies go into the reactor. Lead time through the full chain is measured in years.

Conversion. Three meaningful providers globally: Orano (Comurhex-II in France), Cameco (Port Hope in Canada), Rosatom (Seversk and Angarsk in Russia). The American facility, ConverDyn's Metropolis Works in Illinois, shut down in 2017 and restarted in 2023. During the downtime, ConverDyn lost experienced operators who had taken jobs at refineries and chemical plants along the Gulf Coast where pay was better and job security was not tied to a commodity trading in the teens. Restarting required retraining, equipment refurbishment, and NRC regulatory re-engagement.

Enrichment. About 40% of global capacity belongs to Rosatom. Urenco (jointly owned by the UK, Netherlands, and Germany) operates the largest Western enrichment enterprise, with plants in those three countries plus a facility in New Mexico. Orano operates the Georges Besse II plant in France. That is the full list of Western enrichment capacity.

TENEX, Rosatom's commercial enrichment arm, priced its SWU consistently below Western competitors for decades. According to Euratom Supply Agency data, Russian-origin enrichment services supplied about 25%-30% of EU reactor needs as recently as 2021. The U.S. imported roughly 20%-25% of its enrichment from Russia before the ban. TENEX pricing was low enough that any proposed expansion by Urenco or Orano faced an IRR problem: why would a bank finance new centrifuge capacity when the Russian incumbent could undercut the resulting product price? This dynamic persisted through the Crimea annexation in 2014, through the Salisbury poisoning in 2018, through the initial phase of the Ukraine war. The ban on Russian uranium imports that Biden signed in 2024 (the Prohibiting Russian Uranium Imports Act, part of a broader appropriations package) was the first legislative disruption to this pattern, and it included waiver provisions for existing contracts running through 2027.

The grade distribution in uranium is the most extreme of any major commodity. Athabasca Basin deposits in Saskatchewan grade above 20% U₃O₈ at the top end. Olympic Dam in South Australia grades around 0.03%. The global average across all producing mines is roughly 0.1%. There is nothing comparable in copper or gold where the ratio between the best and worst producing deposits is maybe 10:1 or 20:1, not 600:1.

Cameco's decision to shut McArthur River in 2018 shaped the uranium market for the next four years. I want to spend some time on this because the dynamics involved illuminate how a concentrated commodity market behaves differently from a competitive one.

McArthur River's cash operating cost was in the mid-teens per pound. Spot uranium was in the low-to-mid twenties. The mine was marginally profitable. Cameco chose to shut it and fulfill its delivery contracts from inventory and market purchases instead. The company's 2018 annual MD&A filing describes this as a response to market conditions. Read between the lines and the logic is straightforward: Cameco held enough inventory to honor its contracts. By withdrawing the single largest source of low-cost supply from the market (McArthur River produced roughly 18 million pounds per year at peak), Cameco accelerated a supply tightening that it expected to push prices up over time, benefiting its remaining contract portfolio and the eventual restart of the mine at higher prices. The mine restarted in 2022 as prices recovered.

There is no cartel here. There was no coordination with Kazatomprom or Orano. A single company with the lowest-cost asset decided unilaterally that the market could be more profitably served by withholding supply than by selling it. Whether this is admirable commercial discipline or something more problematic depends on your perspective. The point is that it was possible because the market is concentrated enough for one producer's shutdown decision to meaningfully alter the global supply balance.

Kazakhstan is 43% of world production. All of it ISR. The geology cooperated beautifully: extensive roll-front sandstone deposits at shallow depths across the Syr Darya and Chu-Sarysu basins, amenable to acid leach ISR. Kazatomprom built out production to over 20,000 tonnes per year, more than triple its output in the early 2000s.

The sulfuric acid constraint. Kazatomprom's ISR operations use dilute sulfuric acid. The country's domestic acid production capacity is insufficient. In its Q3 2023 investor presentation, Kazatomprom explicitly cited "difficulties with the timely supply of certain key materials, including sulfuric acid" as a factor in its decision to revise 2024 production guidance downward by about 10%. The global sulfuric acid market is coupled to phosphate fertilizer production (phosphoric acid manufacturing is the largest single use of sulfuric acid) and to copper smelting (acid is a byproduct). Neither of those industries adjusts its output based on what uranium producers need. Kazakh uranium production is a price-taker in the sulfuric acid market and has no ability to influence the availability of its critical input.

I could write a long section about permitting timelines, exploration spending, and the workforce gap, but the essential point is simpler than most analysis makes it. There are exactly two uranium producers in the world with the demonstrated ability to bring substantial new production online within five years: Kazatomprom (by developing new wellfields and deposits within its existing ISR infrastructure) and Cameco (by ramping McArthur River and Cigar Lake back to full capacity). Everyone else is on a ten-to-fifteen year timeline at best, including Orano, which faces the Niger situation, and every greenfield project in Africa, Australia, and the Americas.

The bottleneck is not geological. There is plenty of uranium in the ground. The IAEA/NEA Red Book (Uranium: Resources, Production and Demand, published every two years) consistently reports identified resources sufficient for well over a century of consumption at current rates. The bottleneck is the time and political capital required to convert resources in the ground into producing mines.

Exploration spending data from the IAEA shows 2023 expenditures at roughly one-third of the 2007 peak. The capital market's reluctance to fund uranium exploration, even at current prices, is a residual effect of the 2007-2011 boom-bust. Paladin Energy, Forsys Metals, UraMin, Mega Uranium: a cohort of companies that raised hundreds of millions, found nothing commercial, and incinerated investor capital. That cohort defined a generation of mining fund managers' priors about uranium, and those priors have not updated as fast as the commodity price.

The Megatons to Megawatts program supplied about 24,000 tonnes of natural uranium equivalent per year to the U.S. market from 1993 to 2013. Half of American reactor fuel for two decades came from diluted Russian warheads. When the program ended, that volume vanished from the supply stack. The price impact was delayed by several years because of the post-Fukushima demand collapse and the inventory overhang it created.

Tails re-enrichment. This gets deep into enrichment economics and most generalist coverage skips it entirely, which is a mistake because the volumes involved are large enough to matter.

When a centrifuge plant enriches natural uranium (0.711% U-235) to reactor grade (say 4.5% U-235), it produces a waste stream of depleted uranium, the tails, with some residual U-235 content. The level of residual U-235 left in the tails is the tails assay. It is an economic variable, not a physical constant. If enrichment capacity is abundant and cheap, the operator sets a low tails assay (say 0.20%), extracts more U-235 from each kilogram of feed, and uses more SWU. If enrichment capacity is scarce or expensive, the operator sets a higher tails assay (say 0.30% or 0.35%), uses less SWU, and consumes more natural uranium feed to produce the same amount of enriched product. This SWU-feed trade-off means that enrichment market conditions directly influence natural uranium demand, creating a variable that most simplified supply-demand models ignore.

Russia exploited this mechanism more aggressively than anyone else. With excess centrifuge capacity left over from weapons programs and not fully utilized by commercial orders, Rosatom re-enriched depleted uranium tails that had accumulated at its facilities over decades, some from Russian operations and some stored on behalf of Western and Japanese customers. The tails contained 0.2% to 0.3% U-235. Running them through centrifuges produced a product indistinguishable from natural uranium at 0.711% U-235, which was then sold into the global market. The feedstock was material that the original owners had classified as waste or liability. Rosatom's input cost was effectively the electricity to run the centrifuges. The output sold at natural uranium market prices. According to estimates published by UxC and discussed at WNA Symposium presentations, this tails re-enrichment activity may have contributed 5,000 to 10,000 tonnes of natural uranium equivalent to global supply per year during its peak. Against a mine-production total of 59,000 tonnes and reactor demand of 66,000 tonnes, that is a significant fraction of the gap.

UxC publishes a weekly spot price indicator. TradeTech publishes a competing one. These are survey-based prices derived from broker reports of concluded and pending transactions. They are not exchange-traded prices. There is no uranium futures contract on any major exchange (the NYMEX contract launched in 2007 died from lack of liquidity).

The spot market handles about 10% to 15% of total uranium transaction volume. TradeTech's 2023 Nuclear Market Review reported that approximately 85 million pounds changed hands in the spot market that year, against total reactor requirements of roughly 170 million pounds. The rest moved through long-term contracts whose prices are bilaterally negotiated and not publicly reported.

The participants in the spot market skew heavily toward intermediaries. Trading houses with physical uranium operations (Traxys has been the most active), producers selling excess inventory or managing contract portfolios, and financial funds. Sprott Physical Uranium Trust has been the dominant financial buyer, purchasing physical yellowcake in the spot market and holding it in licensed storage. Sprott's 2023 annual report disclosed holdings of over 60 million pounds of U₃O₈. At peak purchasing rates in late 2021 and 2022, Sprott's buying represented over 30% of spot market turnover. Yellow Cake plc in London holds a smaller but still significant physical position.

Physical delivery in the uranium market is a warehouse receipt transfer on the books of a licensed storage facility. The drums do not move. A lot of uranium can change nominal ownership five times in a month without a forklift touching it. The ratio of financial turnover to physical movement is extremely high.

A utility fuel procurement team operates under a mandate that has no equivalent in other commodity markets: supply must not be interrupted under any circumstances. The cost of an unplanned shutdown due to fuel unavailability is catastrophic in financial and reputational terms. Replacement power costs on the wholesale market run to millions per day. Regulatory scrutiny following a fuel-related outage is intense. The procurement manager who lets this happen does not get a second chance.

Given this mandate, utility buying behavior is systematically biased toward over-procurement and over-contracting. Utilities maintain 2 to 3 years of fuel inventory. They diversify across multiple suppliers. They sign contracts years before the fuel is needed. When the market tightens, they contract more and faster, increasing their coverage ratio. They do not reduce purchases when prices rise because their inventory minimums are non-negotiable.

Uranium's share of nuclear generation cost is small, 5% to 8% depending on the contract vintage. The Lazard Levelized Cost of Energy Analysis, which is the most widely referenced cost benchmarking in the power sector, breaks this out. Fuel cost is a small fraction of the total. A $50 per pound increase in uranium prices adds less than half a cent per kilowatt-hour to the busbar cost. Procurement managers know this number and it shapes their price sensitivity, which is extremely low.

In 2024, multiple Western utilities started including origin clauses in new term contracts. The practical effect is to exclude Russian and Russian-affiliated uranium from their supply chain. TradeTech reported in its Q1 2025 market review that the volume of non-Russian-origin term contracting had increased sharply and that price premiums for guaranteed Western-origin material were widening. If this fragmentation persists, the effective supply available to Western utilities is smaller than the global total, and the tightness that Western buyers face is greater than aggregate global numbers suggest.

Lower fuel efficiency per MWh than large reactors due to neutron economy effects at small core sizes. Some designs require HALEU at 5%-20% enrichment, which has no commercial production infrastructure outside Russia. Centrus Energy's demonstration cascade in Piketon, Ohio, is the only non-Russian HALEU production facility and its output is measured in kilograms, not tonnes.

NuScale's 2023 project cancellation. China's Linglong One operating as a state-funded demonstration. Russia's Akademik Lomonosov powering a remote Arctic mining town. These are the data points that exist. Everything else in the SMR space is projections and promises.

For uranium demand through 2035, SMR is immaterial. China's large-reactor construction pipeline, over 25 units under construction totaling close to 30GW, is where the incremental demand sits. U.S. and European license extensions pushing operating lives from 60 toward 80 years are the second major source. Japanese restarts are the third.

Forty-three percent of global output. Landlocked. Export route through Russian rail to St. Petersburg. JV equity stakes held by Chinese and Russian state entities with contractual offtake rights.

The Chinese equity positions in Kazakh JVs are the part that gets insufficient scrutiny. CNNC and CGN hold stakes in multiple Kazatomprom JVs including operations in the Chu-Sarysu basin. These JV agreements typically include provisions for the Chinese partner to take delivery of a share of production proportional to its equity stake. That production goes directly into the Chinese fuel cycle. It does not appear on the open market. When Kazatomprom reports total national production of 21,000-plus tonnes, some fraction of that output is pre-committed to Chinese offtake and is not available to Western buyers at any price.

How large is that fraction? Kazatomprom's annual report lists its JV partners but does not disclose the specific offtake allocation formulas. Chinese company filings in Mandarin are more forthcoming in some cases but not consistently. Analysts at UxC and TradeTech attempt to estimate the committed volumes but acknowledge significant uncertainty. The range of estimates for Chinese pre-committed offtake from Kazakh JVs that circulates in industry discussions is wide enough to be unhelpful for precise modeling.

Over 25 reactors under construction. Close to 30GW. Commissioning timeline extends from now through 2030 and beyond. A second wave of approvals in 2022 and 2023 added further units to the pipeline. Credible projections from the China Nuclear Energy Association and from Western consultancies like Wood Mackenzie put the Chinese fleet at 100 to 150 operating reactors by the mid-2030s.

Import dependence above 70%. Domestic uranium production covers less than 30% of current requirements. CNNC operates mines in Xinjiang and Inner Mongolia and is investing in ISR technology development for domestic deposits. Even with aggressive domestic expansion, the gap between domestic supply and reactor demand will widen as the fleet grows.

For anyone modeling the uranium market over the next decade, the China variable is the one that keeps you up at night. The demand is large, growing, and relatively predictable from reactor count data. The supply-side behavior (stockpiling pace, contract coverage ratios, willingness to pay above spot for secured volumes) is where the opacity lies, and it is exactly this behavior that determines the residual supply available to everyone else.

ISR has a minimal surface footprint. Modern conventional mining operates under environmental standards that, in jurisdictions like Saskatchewan and Australia, are rigorous and well-enforced.

The Cold War legacy is a different matter. The EPA's listing of abandoned uranium mine sites on Navajo Nation land runs to over 500 locations. The German federal government's WISMUT GmbH, the entity managing remediation of the former East German uranium mining complex, has spent over 6 billion euros since reunification and the work continues. These are not hypothetical risks. They are documented contamination sites with ongoing health consequences.

The completion rate of announced nuclear projects is the determining variable. According to IAEA PRIS data, historically less than half of reactor projects that receive a construction start announcement reach commercial operation. The rest are delayed indefinitely, suspended, or cancelled.

China's completion rate is close to 100% for the current construction cycle. The institutional structure is centralized, the regulatory process is streamlined relative to Western systems, and the political commitment to nuclear expansion has survived multiple changes in leadership emphasis. If China builds everything currently in its pipeline plus a substantial fraction of what is planned, Chinese uranium demand growth alone will tighten the global market materially through the 2030s.

Western completion rates are the uncertainty. If the U.S., UK, France, Poland, Czech Republic, and others follow through on stated nuclear ambitions and find ways to build at reasonable cost and schedule, the combined demand increment, layered on top of China and reactor life extensions, produces a supply deficit that persists for fifteen to twenty years and overwhelms the capacity of existing mines and processing facilities. If Western projects stall in the planning and permitting phase as they often have historically, the tightness is less severe and shorter in duration.

Supply chain rebuilding lags demand growth under either scenario. This is the one structural feature of the uranium market that holds regardless of how the demand side plays out. Enrichment capacity expansion takes close to a decade. Mine permitting in Western jurisdictions takes a decade or more. HALEU production starts from close to zero. The experienced workforce that built and operated the last generation of uranium facilities has thinned out over fifteen years of contraction. New graduates from nuclear engineering programs are not choosing uranium mining and processing as a career path when technology and renewables offer higher compensation and better perceived career trajectories. The human capital constraint may turn out to be the most persistent of all, because it cannot be solved by throwing money at it on a short timeline. Training a centrifuge operator or a uranium hydrometallurgist takes years, not months, and there is no shortcut.