How the Atomic Revival of the AI Era Concentrated the World’s Uranium in Central Asia—Just as Access to It Was Captured at the Mine, the Fuel Cycle, and the Rail

ZeitShift Intelligence | June 2026

KEY THESIS

This is the companion to our analysis of the rare-earth control plane, and it tells the same structural story with a different element, a different geography, and a different lever. In rare earths, the binding constraint was the furnace—the refining monopoly China spent thirty years building. In uranium, the constraint is not a single monopoly but a capture in progress, and it is happening at three layers at once. The first is the mine: the AI-driven nuclear renaissance has revived uranium demand at the precise moment that the two Central Asian states which now anchor global supply—Kazakhstan, the unambiguous world leader, and Uzbekistan, the fast-rising number six—are tightening foreign access to the ground. Kazakhstan’s December 2025 subsoil law gives its state miner a minimum 75 percent of any new joint venture and up to 90 percent on extension; within weeks, the last significant Western explorer in the country packed up three drill rigs and left. The second layer is the fuel cycle: even a state that owns its uranium does not own its fuel, because conversion and enrichment remain concentrated elsewhere, and Central Asia’s ore has historically reached the West only by transiting the very country the entire post-2022 order is trying to route around. The third layer is the rail. The $4.7 billion China–Kyrgyzstan–Uzbekistan railway, financed to close in December 2025 and now under construction, gives landlocked Central Asian resources their first direct artery to China—and, at a small Kyrgyz station called Makmal, it performs a literal break of gauge, switching from the 1,520-millimeter track the Soviet Union laid across the region to the 1,435-millimeter standard China uses. That physical re-gauging is the whole story in miniature: the fuel of the coming nuclear century is concentrating in Central Asia, and the question of who controls it is being answered not at the deposit, where the West still imagines the contest lives, but at the access terms and the track. China is laying the track. The West is sending trade representatives. The structural observation is that, as in rare earths, ownership of the resource and control of the resource are different things—and control is migrating to whoever holds the chokepoint, which in uranium is no longer only the enrichment plant but the entry permit and the rail gauge.

A railway changes gauge where one empire’s standard ends and another’s begins. The break of gauge at Makmal is not an engineering footnote. It is the border of a new order, drawn in steel.

I. Three Signals and the Demand That Lit Them

In our rare-earth analysis we argued that the most valuable intelligence is often a set of events that every desk files separately and no one connects. The uranium story arrived the same way: as three developments, months apart, in three different categories, that describe a single capture.

Signal one, the mine. In April 2026, Uzbekistan’s state uranium enterprise Navoiyuran announced commercial production at its Qizilkok deposit in the Kyzylkum Desert—one node in a national program that lifted the country’s output to roughly 7,000 tonnes of natural uranium in 2025, a 35 percent jump in a single year, and made Navoiyuran, by its own accounting, the sixth-largest uranium producer on earth. Tashkent has authorized four more deposits in 2026 and set a target above 7,200 tonnes by 2030. A country most of the uranium market could not have placed on a map five years ago is now a top-tier supplier, and it is scaling deliberately.

Signal two, the access. On December 24, 2025, Kazakhstan’s president signed amendments to the country’s Subsoil Use Code giving the state miner Kazatomprom—the world’s largest uranium producer—priority rights over exploration licenses and a minimum 75 percent stake in new joint ventures, rising toward 90 percent on extension. Two days earlier, the Toronto-listed explorer Laramide Resources had received its final permits to drill a 15,000-meter program in the Chu-Sarysu Basin, adjacent to the country’s flagship mines. It never turned a bit. By March 2026 it had abandoned the project and left Kazakhstan, with three rigs and full crews on standby that never deployed. One of the last Western companies exploring for uranium in the world’s most important uranium country walked away because the terms of access had been rewritten underneath it.

Signal three, the rail. On December 16, 2025, in Bishkek, a joint venture owned 51 percent by China signed the loan agreement that financed the long-delayed China–Kyrgyzstan–Uzbekistan railway—$4.7 billion, more than half of it a 35-year Chinese state loan—closing three decades of stalled negotiation and clearing the way for a line that will give Central Asia its first direct rail link to China that does not run through Kazakhstan, and route freight to Europe in a way that does not run through Russia.

Three signals: a mine scaling, an access door closing to the West, a rail being laid by China. What lit all three is a fourth fact that has nothing to do with Central Asia and everything to do with why Central Asia suddenly matters—the return of nuclear power as a strategic industry, driven by the one demand source no forecaster modeled five years ago: artificial intelligence.

II. The Nuclear Renaissance Has a Materials Problem

For two decades after Fukushima, uranium was a stranded asset class—oversupplied, politically radioactive, written off by Western capital as the fuel of a declining industry. That consensus has inverted, and the agent of inversion is the data center.

The AI buildout needs firm, dense, around-the-clock power on a scale and timeline that intermittent renewables cannot supply alone, and the hyperscalers have concluded—publicly, with signed offtake agreements and small-modular-reactor commitments—that nuclear is the answer. The macro numbers describe a genuine inflection. The World Nuclear Association’s reference scenario has global reactor capacity rising from roughly 372 gigawatts today to about 449 by 2030 and 746 by 2040, an average growth rate above 5 percent a year, with some 60 gigawatts of additional new builds layered on versus the prior forecast—concentrated in East and South Asia and in newcomer states. The fuel implication is the part that matters here: annual demand for natural uranium, in that scenario, climbs from roughly 67,000 tonnes today to over 150,000 tonnes a year by 2040. The industry is being asked, in other words, to roughly double primary supply inside fifteen years, after fifteen years of being told not to invest.

What makes this inflection different from the false dawns of the past is that it is no longer a forecast; it is a contract book. As of mid-2026, the four hyperscalers had signed roughly thirteen nuclear deals committing close to 9.8 gigawatts of capacity specifically to power AI infrastructure. Microsoft anchored the turn in 2024 with a 20-year, $16 billion power-purchase agreement to restart Three Mile Island Unit 1—rebranded the Crane Clean Energy Center, 835 megawatts—whose first electrons are expected around 2027, making it the first mover precisely because it is a restart rather than a new build. Amazon put roughly $700 million into the small-modular-reactor developer X-energy for up to twelve Xe-100 units, targeting some 5 gigawatts by 2039, and committed on the order of $20 billion to convert the Susquehanna site into an AI campus. Google signed the first corporate SMR offtake agreement with Kairos Power for 500 megawatts by 2035. And Meta committed the largest block—up to 6.6 gigawatts by 2035 across Vistra, Oklo, TerraPower, and Constellation, including a TerraPower Natrium agreement that could scale to eight reactors and 2.8 gigawatts—while issuing a separate request for proposals for one to four gigawatts more. Around these sit Westinghouse’s plan for ten pre-licensed AP1000 reactors by 2030, the $500 billion Stargate data-center project, and a U.S. policy turn—four nuclear executive orders in May 2025, an accelerated NuScale design approval—engineered to compress the timelines.

HyperscalerPartner / assetCommitmentTypeFirst power
MicrosoftConstellation — TMI Unit 1 (Crane)835 MW, $16B 20-yr PPARestart~2027
AmazonX-energy (+ Susquehanna campus)~5 GW by 2039; $700M + ~$20BSMR + restart~2030s
GoogleKairos Power (Hermes 2)500 MW by 2035Gen IV SMR~2030
MetaVistra, Oklo, TerraPower, Constellationup to 6.6 GW by 2035Mixed + SMR2032–2035
Aggregate~13 announced deals~9.8 GW committed—2027 onward

The significance for uranium is not the gigawatts in isolation but what they imply for fuel. These reactors, restarts, and SMRs all consume enriched uranium, and several of the advanced designs require high-assay low-enriched uranium that the West currently cannot produce at scale. The demand signal that was once a line on a World Nuclear Association chart is now capital committed by the most cash-rich companies on earth—which means the pressure on the fuel chain is no longer hypothetical, and the question of who controls that chain is no longer academic.

This is the same shape of problem we identified in rare earths and copper: a demand curve bent sharply upward by AI, colliding with a supply base that was deliberately starved of capital during the years it should have been expanding. And it produces the same structural vulnerability. When demand for a strategic input rises faster than supply can be reorganized, whoever controls the chokepoint in the supply chain acquires leverage that is disproportionate to their share of the raw resource. In rare earths the chokepoint was processing. In uranium it is more complicated, because the uranium fuel chain has two chokepoints—and Central Asia, for all its new dominance in mining, sits upstream of both.

That is the tension the rest of this report unpacks. The world is rushing back to nuclear power to feed the AI economy. The uranium to fuel it is concentrating in two Central Asian states. And the control of that uranium—who gets to mine it, who gets to move it, who gets to turn it into fuel—is being decided right now, at terms and on infrastructure that are tilting decisively east.

III. The Center of Gravity Moved to the Steppe

To understand why a rewritten mining code in Astana and a railway through the Kyrgyz mountains are global events, you have to see how completely the geography of uranium supply has concentrated.

Kazakhstan is not merely the largest uranium producer; it is dominant in a way few commodities are. Its state champion Kazatomprom alone accounts for something close to a quarter of global primary supply, and Kazakhstan’s mines—counting the joint ventures that operate on its soil—produce on the order of 40 percent of the world’s uranium. In 2025 the country set a record at roughly 25,800 tonnes. The dominance rests on geology and method: Kazakhstan’s deposits are suited to in-situ recovery, in which a solution is pumped underground to dissolve the uranium and bring it to the surface—the lowest-cost extraction technique in the industry, requiring no open pit and no underground mine. Kazakhstan can produce uranium more cheaply than almost anyone, which is precisely why everyone wants in, and why its terms of access matter so much.

Uzbekistan is the rising complement. Long invisible to the conventional uranium narrative—which fixed on Kazakhstan, Canada, and Namibia—it produced roughly 7,000 tonnes in 2025 and now ranks as the world’s fifth-largest producing country, with its state enterprise Navoiyuran the sixth-largest producer by company. Navoiyuran was spun out of a Soviet-era conglomerate only in 2022; it uses the same low-cost in-situ method as its Kazakh neighbor; it reports a reserve base around 96,600 tonnes across more than forty deposits; and it is scaling fast, with commercial production newly online at Qizilkok and four further deposits slated for development in 2026. Crucially, uranium is the spearhead of a broader resource push: Uzbekistan launched a $2.6 billion initiative spanning 76 mining projects across 28 different elements—many of them the same critical minerals the West is scrambling to source outside China. The country is positioning itself not just as a uranium supplier but as a diversified strategic-minerals jurisdiction.

Kazakhstan (Kazatomprom)Uzbekistan (Navoiyuran)
Global rank#1 producer (~40% with JVs)#5 country / #6 company
2025 output~25,800 t (record)~7,000 t (+35% YoY)
MethodIn-situ recovery (lowest cost)In-situ recovery
Access trendTightening: ≥75% state stake in new JVsState-directed; foreign JVs (Orano, CNNC, JOGMEC)
2030 trajectory”Value over volume”; managedTarget >7,200 t; four new deposits

Two adjacent states, using the same cheap method, now anchor the upstream of the entire nuclear fuel chain. This concentration is the precondition for everything that follows. A diffuse resource base cannot be captured; a concentrated one invites it. And both states have read the same lesson the rest of the resource-producing world has absorbed—the lesson we traced in the rare-earth analysis, that the value and the leverage lie in controlling access and the downstream, not in shipping raw ore cheaply. The difference is that in 2025 and 2026, Central Asia stopped studying the lesson and started applying it.

IV. Mining Was Never the Constraint

Here the uranium story diverges from the rare-earth one in a way that makes it more, not less, dangerous for the West—because uranium has not one chokepoint but two, and Central Asia’s new mining dominance sits upstream of both.

Natural uranium is nearly useless as it comes out of the ground. To become reactor fuel it must be converted into uranium hexafluoride gas, then enriched to raise the concentration of the fissile isotope, then fabricated into fuel assemblies. Conversion and enrichment are the hard, capital-intensive, technically guarded stages—the uranium equivalent of the rare-earth furnace—and they are concentrated in a small number of facilities worldwide. For the Western fuel chain, the uncomfortable fact is that a large share of global commercial enrichment capacity sits in Russia. This is the part of the chain the entire post-2022 order has been trying to de-risk, and it has proven the stickiest, because enrichment plants take the better part of a decade and an enormous amount of capital and political will to build, and the West spent thirty years letting that capacity migrate east on the logic that fuel was a solved problem.

So owning the mine is not owning the fuel. A Western utility that buys Kazakh or Uzbek uranium still faces the conversion-and-enrichment bottleneck downstream—and, until very recently, faced an even more basic problem upstream of it: getting the uranium out of Central Asia at all. The region is landlocked. For decades, the fastest and cheapest route for Kazakh uranium to reach Western buyers ran north and west through Russia, to the port of St. Petersburg. That route was a quiet dependency nobody priced—until it wasn’t. When relations between Russia and the West ruptured, the Canadian producer Cameco, which operates one of the largest joint ventures in Kazakhstan, suspended deliveries of Kazakh uranium while it searched for a route that avoided Russian territory. The mine was fine. The logistics were captured.

This is the structural insight that connects uranium to everything we have written about the physical layer: the constraint is almost never the deposit. It is the furnace, or the enrichment plant, or the port, or the rail—the step that converts a raw material into something usable, or moves it from where it is to where it is needed. The West’s mental model of the uranium contest is geological: secure the mines, secure the supply. But the mine is the one part the West can least affect, because the mines are in Central Asia and the access terms are being rewritten by the states that own them. The parts the contest actually turns on—enrichment, and now the rail that moves the ore—are exactly the parts tilting east. Which brings us to the access layer, where the capture is most visible and most recent.

V. The Capture of Access

For most of the post-Soviet period, Central Asia’s resource jurisdictions operated on a model the West found comfortable: foreign companies brought capital and technology, took equity in joint ventures, and exported the output. That model is being dismantled in real time, and uranium is where the dismantling is sharpest.

Kazakhstan moved first and most explicitly. The Subsoil Use Code amendments signed in late December 2025 did three things at once. They gave Kazatomprom priority rights to obtain exploration licenses in prospective areas. They set a floor of 75 percent state ownership in new uranium joint ventures—and up to 90 percent when an existing contract is extended, which can compress a foreign partner’s stake to a tenth. And they provided that if another party discovers uranium, discovery confers no priority right to produce it; the block can revert to the state. The combined effect, as one mining analyst put it bluntly, is to discourage essentially any foreign company from exploring in Kazakhstan at all. The Laramide exit was the immediate, concrete proof: permits in hand, rigs ready, program funded—and then a signature in Astana made the economics impossible, and a Western explorer walked. Kazakhstan paired this with a deliberate pivot to “value over volume,” managing output to support prices rather than maximize tonnage—the posture of a producer that knows it holds a scarce, strategic asset and intends to price it accordingly.

Uzbekistan’s version is different in form, identical in direction. Tashkent never opened to Western majors the way Kazakhstan did; its uranium has always run through the state enterprise Navoiyuran, and the post-2022 expansion has been state-directed from the start. Foreign partners are admitted as minority technical collaborators—France’s Orano in the Nurlikum joint venture, Japan’s JOGMEC and Itochu in exploration—on terms the state sets. And the most consequential partnership of the recent period was not with a Western firm at all: in September 2024, Navoiyuran signed a trilateral agreement with China National Uranium Corporation, a subsidiary of the China National Nuclear Corporation, covering joint geological exploration and a shared scientific-technical center. The pattern across both states is the same. Access to Central Asian uranium is being re-priced upward and re-allocated—away from the open, foreign-equity model the West relied on, toward state control with a strong eastern tilt.

This is not, in itself, a hostile act, and it is important to read it without the reflex to moralize. It is what resource-rich states do once they understand the leverage they hold—the same dynamic of producer states demanding the furnace as the price of the ore that we documented across Brazil, Indonesia, and India in the rare-earth analysis. Kazakhstan and Uzbekistan are doing the rational thing: capturing more of the value of an asset that just became scarce and strategic. The consequence for the West is simply that the entry point is narrowing exactly as the need is widening. And the actor positioned to walk through the narrowing door—because it offers not just to buy the ore but to finance the mine, build the reactor, and lay the rail—is China. The second half of this report is about the track.

PART 2: THE TRACK AND THE CONTEST

VI. Break of Gauge

Every railway that crosses the old Soviet frontier has to solve a physical problem, and the solution is a metaphor for the entire geopolitics of the region.

The Soviet Union built its rail network on a gauge of 1,520 millimeters—deliberately wider than the 1,435-millimeter standard used in China, Europe, and most of the world, a choice that was as much strategic as technical: a different gauge is a passive defense, slowing any army or economy that tries to roll across the border. A century later, that gauge difference is still there, and it still has to be reconciled every time goods move between the Chinese system and the post-Soviet one. The reconciliation happens at a “break of gauge”—a transshipment point where cargo is lifted from one set of bogies to another, or where the track itself changes width.

The China–Kyrgyzstan–Uzbekistan railway will perform its break of gauge at a Kyrgyz station called Makmal. West of Makmal, toward Uzbekistan, the line runs on the 1,520-millimeter Soviet gauge. East of it, toward Kashgar, it runs on the 1,435-millimeter Chinese standard. The railway is, quite literally, the seam where one empire’s infrastructural standard ends and another’s begins—and the direction of the new construction tells you which way the seam is moving. For thirty years Central Asia’s external rail connections defaulted to the Soviet gauge and the Russian network. The CKU line is the first major artery built to connect the region directly to the Chinese standard instead. The break of gauge at Makmal is where Eurasia is being re-gauged.

The project itself is now real in a way it was not for the three decades it spent on paper. The route runs roughly 523 kilometers—about 213 in China, 260 across the Kyrgyz mountains, and 50 in Uzbekistan—from Kashgar through the Torugart Pass to Andijan in the Ferghana Valley. It is among the most demanding engineering projects in the region: some 50 bridges and 29 tunnels, together about 40 percent of the entire length. The cost is put at $4.7 billion, with China providing roughly half as a 35-year concessional loan and holding 51 percent of the joint venture; Kyrgyzstan and Uzbekistan split the rest. Financing closed in Bishkek in December 2025; construction had ceremonially begun a year earlier at Jalal-Abad, after the Ferghana Valley states settled the border disputes that had blocked the route. Completion is targeted for 2029 to 2030. At full capacity the line is designed to move up to 15 million tonnes of cargo a year and to cut transit time between China and Europe by seven to ten days against the northern route through Russia.

For Kyrgyzstan—a country whose entire rail network amounts to about 424 kilometers of disconnected Soviet-era track, and which has never had a direct rail link to China—the line is the largest infrastructure project in its history. Its president has framed the stakes in exactly the terms that matter to this analysis: a country that has been, in his words, a “dead-end” between Europe and China will become a transit power. That is the prize the corridor confers, and it is why three landlocked states agreed to take on Chinese debt to build it. It is also why the line’s strategic meaning—stated plainly in the regional analysis, and unstated but obvious in Beijing—is that it creates a Eurasian trade artery that bypasses Russia, and gives Central Asia’s resources, uranium among them, a path to market that runs east instead of north.

VII. China Is Laying the Track

The railway is not a standalone gambit. It is the connective tissue of an integrated position that spans the entire uranium chain—and integration is the point.

Consider what China is assembling in Central Asia, layer by layer. At the mine: equity and offtake, from the historical CGN purchase agreements through the 2024 trilateral exploration deal between Navoiyuran and China National Uranium Corporation, positioning Chinese entities inside the upstream of both Kazakh and Uzbek production. At the reactor: China is building nuclear plants in Kazakhstan—reporting around the second Almaty-region project has noted China gaining the edge—which converts a uranium supplier into a captive customer and an operating partner, binding the relationship at both ends of the fuel chain. At the rail: the CKU line, plus the existing China–Kazakhstan crossings that already moved a record 32 million tonnes of freight in 2024, plus the Kashgar free-trade zone built to turn Xinjiang into the western gateway. And around all of it: the financing—Chinese state loans on 35-year terms—that makes the infrastructure happen on China’s timeline and ties the borrowers into a long relationship.

This is the move the West structurally cannot match with the tools it is using, and it is worth naming precisely why. China is not offering Central Asia a purchase order. It is offering a system: finance the mine, build the reactor that consumes the output, lay the rail that moves it, and settle the trade—increasingly in yuan—through a zone it controls. Each piece reinforces the others. A reactor in Kazakhstan creates demand that justifies the mine; the rail makes the mine’s output movable; the loan makes the rail buildable; the offtake makes the loan serviceable. It is the same integrated logic that built China’s processing dominance in rare earths, applied to a different element and extended from the factory to the railhead. The West, by contrast, is offering the parts—a purchase agreement here, a diplomatic visit there—without the connecting infrastructure that turns parts into leverage.

The result is that even where Western firms remain present—Cameco in its Kazakh joint venture, Orano in Uzbekistan—they increasingly operate inside a system whose financing, logistics, and downstream are being built by someone else. Presence is not the same as position. You can hold equity in a mine whose output moves on a rail you did not build, to a reactor you do not operate, financed by a loan you did not extend, and discover that your equity buys you less control than you assumed. That is the situation the integrated Chinese play is engineering: a Central Asian uranium complex in which Western firms are minority tenants and China is the landlord, the banker, the customer, and the railway.

VIII. The West Is Courting, Not Building

It would be wrong to say the West is absent. It is, in fact, unusually active in Central Asia in 2026—more active than it has been in years. The problem is the form the activity takes.

The American re-engagement has been concrete and recent. Uzbekistan and the United States launched a joint investment platform—explicitly tagged to critical minerals—announced negotiations toward a comprehensive trade pact, and Tashkent proposed a special economic zone for U.S. companies; President Mirziyoyev received the U.S. Trade Representative, the same official who convened the critical-minerals coalition we covered in the rare-earth analysis. The European Union signed a strategic-partnership framework with Kazakhstan built around the Middle Corridor, opened a European Investment Bank office in Tashkent, and—through Germany—anchored a Volkswagen assembly and logistics project in Uzbekistan. France’s Orano holds its Uzbek joint venture; Japan’s agencies hold exploration licenses. This is real engagement by serious actors, and it reflects a genuine recognition that Central Asia has become strategically central.

But notice the shape of what the West is offering: investment platforms, trade pacts, purchase frameworks, diplomatic partnerships—instruments of access and offtake. What it is conspicuously not offering at comparable scale is the integrated, financed, infrastructure-and-reactor build that China is putting on the table. The West wants to buy the uranium and the minerals; China wants to build the system that produces and moves them. Those are not symmetric bids, and the asymmetry is structural, not a matter of effort or goodwill. Western capital is private, return-seeking, and allergic to 35-year concessional loans into politically complex landlocked states; Chinese capital is state-directed and patient, and treats the loan’s strategic return as part of its yield. A Western trade representative can sign an investment platform in an afternoon. A Western consortium cannot, on the same logic, finance a $4.7 billion mountain railway or build a nuclear plant on terms that compete with Beijing’s.

The asymmetry is clearest laid out layer by layer, because it is not that the West is doing nothing—it is that it is competing on a different and shallower set of layers than the one that decides control.

LayerThe China bidThe Western bid
Mine accessEquity + offtake (CNNC–Navoiyuran trilateral; CGN history)Minority JV stakes (Cameco/Inkai, Orano/Nurlikum), now compressing under new subsoil terms
Finance35-year state concessional loans, strategic returnPrivate, return-seeking capital; quarterly horizon
ReactorBuilding plants in Kazakhstan (captive demand)Domestic builds at home; little new build in-region
Rail / logisticsCKU + China–Kazakhstan crossings + Kashgar FTZCourting the Middle Corridor; financing little of it
Fuel cycleExpanding conversion/enrichment at homeRebuilding from near-zero (Orano TN, Centrus, Urenco)
SettlementIncreasingly yuan, inside a controlled zoneDollar, open market
InstrumentAn integrated systemÀ la carte parts
Time horizonMulti-decade, patientElectoral / earnings cycle

Read down the two columns and the structural point is unavoidable: the Western bid is strong precisely on the layers that confer the least durable control—purchase agreements, minority equity, diplomatic frameworks, all reversible—and weak on the layers that confer the most: the loan, the rail, the reactor, the fuel cycle, the settlement rail. China is bidding on the parts that, once built, cannot easily be un-built.

So the West arrives with the right recognition and the wrong toolkit. It has understood that Central Asia matters; it has not reconciled itself to the kind of state-backed, long-horizon, infrastructure-first commitment that the contest now requires—the same commitment it has been slow to make in domestic rare-earth processing, and for the same reason. The courtship is sincere. It is also, on current form, losing to the bloc that is laying track while the West is signing memoranda. Recognition without the matching instrument is how you come second in a contest you have correctly identified.

IX. The Multi-Vector Hinge

None of this means Central Asia is being absorbed. The most important actors in this story are not Washington or Beijing or Moscow but Astana, Tashkent, and Bishkek—and they are playing a deliberate game that is older than any of the current entrants.

The doctrine has a name in the region: multi-vector foreign policy, the practice of balancing relations among competing great powers so that no single one acquires decisive control. Central Asian states have run this strategy since independence, and the resource-and-corridor moment has made it more lucrative than ever. Kazakhstan ships uranium east to China and west toward Europe, builds reactors with Chinese and Russian and potentially Western partners, and rewrites its subsoil law to extract more value from all of them. Uzbekistan signs a trilateral uranium deal with China’s nuclear corporation and a joint investment platform with the United States in the same season. Kyrgyzstan takes Chinese rail finance while courting Western development banks and reassuring foreign capital. The break of gauge at Makmal is, from Bishkek’s point of view, not a surrender to the Chinese standard but a second connection added alongside the Russian one—optionality, not capture.

This is the genuine analytical complication, and it is worth holding against the report’s own argument. The states at the center are not pawns; they are the ones monetizing the contest, and their interest is precisely in keeping it a contest—in ensuring that China’s integrated bid is balanced by enough Western and Russian and Turkish and Gulf presence that they retain leverage over all of them. A transit power’s value comes from connecting in every direction, not from belonging to one bloc. Kyrgyzstan’s leadership talks openly of becoming a hub that links China to the Middle East and South Asia through the Trans-Afghan corridor as well as to Europe—a node, not a vassal.

But multi-vector balancing has a structural weakness that the integrated Chinese play is designed to exploit, and it is the same weakness that undoes the Western courtship: balance requires comparable counterweights, and infrastructure is not easily counterweighted once it is built. A state can balance purchase agreements against each other freely, because they are reversible. It cannot as easily balance a railway it took thirty-five years of Chinese debt to build, or a reactor China operates, or an enrichment relationship, because these are durable physical facts that bend the flow of goods and money for decades. The multi-vector states can keep the contest open at the level of contracts and diplomacy. At the level of steel and concrete and 35-year loans, the vectors are not equal—and the heaviest vector, right now, is the one laying the track.

X. The Two-Tier Fuel Market

The capture of access does not announce itself as a shortage. It announces itself as a split—the same bifurcation we traced in rare earths, arriving in uranium along a different seam.

A Western utility fueling a reactor in 2026 faces a chain of choices it could ignore five years ago. It can buy uranium from Central Asia at competitive prices—but the cheapest logistics ran through Russia, and the politically safe route does not, so it pays a transit premium or accepts delay. It must then convert and enrich the material—and the secure, non-Russian capacity for that is scarce and commands a premium of its own, while the abundant capacity is the one it is trying to avoid. At every stage, there is now a cheap option that carries strategic exposure and a secure option that costs more, and the gap between them is widening into a structural feature of the market rather than a temporary dislocation. Analysts have begun describing uranium as a “tale of two markets,” and the description is precise: one tier optimized for cost and integrated eastward, one tier optimized for security and paying for the privilege.

The headline price masks the split rather than revealing it. Spot uranium ran to a sixteen-year high near $91 a pound at the end of 2023, then fell back through 2024 to close around $73.50, dropped to roughly $64.50 by early 2025 amid tariff and policy uncertainty, and by mid-2026 sat in a narrow band around $85—lifted by the AI-nuclear demand story but capped because utilities, wary of short-term trade risk since 2022, have leaned on long-term contracts rather than buy spot. But the spot number is the least important figure in the chain, and this is the analytical core of the two-tier market: at the prices utilities actually pay for delivered fuel, only about one-third of the cost is the ex-mine uranium itself. The other two-thirds is enrichment and fabrication, with a sliver for conversion. The value—and therefore the leverage—sits downstream of the mine, in exactly the stages the West has hollowed out.

The numbers there are starker than any yellowcake quote. The United States went from being the world’s dominant uranium enricher to owning essentially zero domestic enrichment capacity; it operates a single foreign-owned enrichment plant rated at 4.9 million separative work units a year against a national requirement near 15 million, and a single conversion plant. Russia supplied close to a quarter of the enriched uranium feeding U.S. reactors—about 3.9 million SWU, roughly 24 percent of demand, in a recent year—which is why Washington banned Russian enriched-uranium imports through 2040 (with waivers only to 2028), why Moscow retaliated with its own export ban, and why the law released $2.7 billion to rebuild domestic enrichment and high-assay capacity from near-scratch. The enrichment price itself softened to about $97.66 per SWU in 2024 from $106.97 the year before, but the relevant fact is not the price—it is that the West must now pay to rebuild a capability it spent thirty years offshoring, while the demand wall arrives.

The contracting data exposes how far behind the security-conscious tier has fallen. Total utility term contracting in 2024 came to about 116 million pounds—down 29 percent year-on-year—and a single Chinese contract accounted for nearly 40 percent of that volume, against annual reactor requirements estimated near 175 million pounds. Western utilities, in other words, are under-replacing their fuel while a single eastern buyer locks in a large block. And the World Nuclear Association notes the deeper pattern plainly: the vertically integrated sovereign nuclear industries of China and Russia have been buying equity in uranium mines abroad—China holds stakes in mines in Niger, Namibia, Kazakhstan, Uzbekistan, and Canada—bypassing the open market entirely. The two-tier market is not a forecast. It is already priced in, in the contracts and the equity stakes, beneath a spot number that tells you almost nothing about who controls the fuel.

Kazakhstan’s posture sharpens the split deliberately. By managing output for “value over volume” and tightening foreign access, the dominant producer is signaling that it intends to price uranium as the scarce strategic asset it has become rather than flood the market—which lifts prices for every buyer, but lifts them most for the security-conscious Western tier that cannot simply accept the integrated Eastern terms. The eastern buyers, meanwhile, are locking in supply through the integrated system described above: equity, offtake, reactors, rail, all settled inside a relationship that prices differently than the open market. The outcome is a fuel market increasingly stratified by alignment, in which what you pay for a kilogram of usable reactor fuel depends less on the spot price of uranium than on which side of the access-and-routing divide you sit.

This is the consequence that should concern Western planners more than any single mine or railway, because it is durable and self-reinforcing. A security premium funds the alternative supply that justifies the premium; an integrated discount deepens the dependency that makes the discount possible. The two tiers pull apart, and the AI-driven demand surge arrives on top of the divergence, raising the stakes of being on the expensive, exposed side. As in rare earths, the unusual and uncomfortable feature is that the strategically necessary tier is also the costlier one—and the cost is not a market accident but the price of having let the cheap, integrated chain be built by someone else.

XI. Steelman: The Capture Won’t Hold

The argument of this report is that control of Central Asian uranium is migrating east through access terms, integrated infrastructure, and the rail. The strongest case against that argument deserves a full hearing, because parts of it are sound.

The railway may underdeliver. The CKU line is one of the most demanding engineering projects in the region—40 percent bridges and tunnels through high mountains—and serious analysts judge that its terrain-limited capacity may make it a feeder route rather than a true alternative to the existing China–Kazakhstan corridor. A single-track, non-electrified line with a break of gauge at Makmal is not a frictionless artery; transshipment and gauge changes impose real costs and delays. If the line moves a fraction of its designed 15 million tonnes, its strategic weight shrinks accordingly, and the “re-gauging of Eurasia” becomes a slower, partial affair than the financing ceremony implied.

Uranium is abundant and the market is liquid. Unlike the heavy rare earths, uranium is not geologically scarce, and primary mining is not the only source: secondary supply, inventories, reprocessing, and a deep long-term contract market give Western utilities buffers that a sudden access shock does not immediately drain. Canada, Australia, Namibia, and a reviving U.S. domestic sector exist outside Central Asia entirely. A two-tier market is uncomfortable, but it is not a cutoff, and high prices are precisely the signal that calls forth non–Central Asian supply.

The producers need the West. Kazakhstan and Uzbekistan are pursuing multi-vector strategies for a reason: they do not want to be captured by China any more than by Russia or the West, and they need Western buyers, capital, and technology to keep their options open. An overreach toward Beijing would cost them the leverage their whole strategy depends on. The access tightening is about extracting value, not about choosing a side—and a value-maximizing producer keeps every customer in the room.

The demand wall may not arrive on schedule. The nuclear-for-AI thesis rests on hyperscaler commitments, and several are SMR designs that have never been commercialized in the United States; the macro buildout itself shows strain, with data-center announcements slowing in late 2025 and grid-delivery limits described as a “bend in the trajectory” by early 2026. If AI demand cools, if SMRs slip, or if the grid simply cannot deliver power on hyperscaler timelines, the uranium demand surge softens and the urgency behind the whole contest eases—and a less urgent contest is one the West can afford to lose more slowly.

These points are real, and a complete analysis grants them: the railway’s capacity is uncertain, uranium is more substitutable and better-buffered than rare earths, the producers need plural customers, and the demand surge could underdeliver. The rebuttal is not that these forces are absent but that they operate at the wrong layer and the wrong speed. The capacity question concedes the direction while disputing the magnitude—a feeder route is still a new eastward artery where none existed, and infrastructure tends to grow into its capacity once built. Abundance and liquidity protect against a cutoff, which is not the claim; the claim is a structural cost premium and a creeping loss of position, against which inventories are a buffer, not a cure. A softer demand wall slows the timeline but does not reverse the access capture or un-pour the rail—and the infrastructure being built now will still be there when demand resumes, because that is what infrastructure does. And the multi-vector hedge, as established, holds at the level of contracts but not at the level of the durable physical infrastructure that is being poured in concrete right now. The steelman is correct that this is not a seizure. It is something slower and harder to reverse: a re-weighting of the whole system toward the bloc that is building it, one loan and one tunnel at a time. The capture does not have to “hold” in the sense of a lock. It only has to accumulate.

XII. Timeline: 2026–2032

The contest will resolve not on a single date but across a window, and the gates are visible.

2026 — The terms reset. Kazakhstan’s subsoil regime takes full effect; further Western explorers reassess or exit in Laramide’s wake. Uzbekistan brings four new deposits toward development and presses its $2.6 billion, 28-element resource push. The U.S.–Uzbekistan investment platform and the EU–Kazakhstan partnership generate announcements; the question is whether any convert into financed infrastructure rather than offtake. CKU tunneling advances. Nuclear-for-AI offtake deals continue to multiply in the West, tightening the demand outlook without yet moving physical supply.

2027–2028 — The build diverges. The gap between the integrated Eastern model and the Western purchase-and-partnership model becomes legible. China-built reactors in Kazakhstan progress; CKU’s hardest tunnel sections determine whether the 2029–2030 target holds. The conversion-and-enrichment bottleneck remains the West’s stickiest problem, with new Western capacity slow to materialize. The two-tier market widens as security premiums on non-Russian-routed, non-Chinese-aligned fuel become a standing feature.

2029–2030 — The track opens. If on schedule, the CKU line begins operating, giving Central Asian resources their first direct, Russia-bypassing rail to China—and the break of gauge at Makmal becomes a working seam rather than a planned one. Reactor capacity continues its climb toward the WNA’s 449-gigawatt 2030 marker; uranium demand pulls further ahead of comfortable supply. Whether the railway functions as artery or feeder is now an empirical fact rather than a forecast.

2030–2032 — The position consolidates. The integrated Chinese complex—mine equity, offtake, reactors, rail, settlement—matures into a durable position across the upstream of the nuclear fuel chain, while the multi-vector states continue to balance at the contract level. The West, on current trajectory, holds residual equity and long-term contracts but not the integrated build, and pays the security premium for the privilege. Independence, in the strong sense, is not on this timeline. The realistic question, as in rare earths, is narrower: whether the West secures enough non–Central Asian, non-Russian-routed, non-Chinese-enriched fuel to power its own nuclear revival without that fuel’s price and availability being set inside a system someone else controls.

The projection, stated plainly: by the early 2030s the contest does not produce a Central Asian bloc-membership, but it does produce a uranium chain whose physical and financial spine—the rail, the reactors, the loans, the enrichment relationship—tilts east, with the West a paying participant rather than an architect.

XIII. Conclusion: The Re-Gauging

Return to the seam at Makmal, because it contains the whole argument.

For a century, the standard gauge across Central Asia was the one the Soviet Union laid—1,520 millimeters, wider than the world’s, a passive assertion of which system the region belonged to. The infrastructure encoded the allegiance; the track was the politics. That is the deepest lesson of the physical layer, the one we keep relearning across every domain: infrastructure has memory. It does not merely move goods; it remembers, for decades, the strategic logic of whoever built it, and it bends the flow of trade and money toward them long after the builder’s moment has passed. The Soviet gauge outlived the Soviet Union by thirty-five years because the steel was still in the ground.

What is happening now is a re-gauging. The China–Kyrgyzstan–Uzbekistan railway lays new standard-gauge track to the Chinese border and performs its break of gauge at Makmal—and in doing so it begins to encode a different allegiance in the steel, one that will bend the region’s trade eastward for the decades the track survives. The uranium is the cargo that makes the re-gauging strategic rather than merely commercial: the fuel of the AI-driven nuclear century, concentrated in two Central Asian states, with its access captured at the mine, its fuel cycle bottlenecked downstream, and now its export artery being built to a new standard by the power that also wants to finance the mines, operate the reactors, and settle the trade.

This is not a moral claim, and it is not a prophecy of conquest. The multi-vector states are not being annexed; they are monetizing a contest they are skilled at keeping open, and uranium is abundant enough and the market liquid enough that no one will be starved of fuel. The claim is narrower and harder to escape: control of a strategic resource is being decided, as it always is, not at the deposit but at the chokepoints around it—and in uranium those chokepoints are now the access permit, the enrichment plant, and the rail gauge, each of which is tilting away from the West at the precise moment the West needs the fuel most. The deposit was never the prize. The prize was the furnace, in rare earths; in uranium it is the fuel cycle and the track. Whoever holds those holds the floor under the nuclear revival, and right now the West is buying access to a floor that someone else is pouring.

The gauge changes at Makmal because that is where one order’s standard ends and another’s begins. The fuel of the next century will cross that seam. The open question is no longer whether the seam exists. It is how far west it moves.

DATA SOURCES

This analysis synthesizes scanner intelligence with primary documentation and industry reporting, cross-referenced for quantitative accuracy.

Scanner Signals (June 2026):

Uranium Production & Reserves:

Access & Regulatory Capture:

Fuel-Cycle & Routing:

Pricing & Contracting:

Demand Context:

CKU Railway:

Western & Allied Engagement:

Cross-Reference Validation:

All quantitative claims are supported by scanner data, publicly available sources, industry-analyst reports, or technically grounded estimates.

Companion Analysis:

ZeitShift Intelligence | Infrastructure has memory. The gauge is the politics.