AI Power Predation and the Uranium Supercycle: Valuation Distortions from Big Tech Nuclear Privatization [EN]

"The only physical limit controlling the expansion rate of artificial intelligence models is not semiconductors, but the power grid. And without the resurrection of overwhelming baseload power known as nuclear energy, there is no physical way on Earth to drive this massive system of intelligence capital."
  — Sam Altman, OpenAI CEO (Remarks at the 2024 Davos Forum / Applied to: 2026 Macro Infrastructure System Analysis)

* The original data and baseline analysis of this macroeconomic shift are available in the Korean report. -> Korean Version

Prologue: A Market Observer's Perspective

This report proves with data the mechanisms of comprehensive restructuring and collapse of the macroeconomic infrastructure system, where the infinite expansion of Intelligence Capital (AI) collides head-on with the physical limits of renewable energy, inevitably forcing global capital to revert to uranium and nuclear power—the heaviest and most strictly controlled real assets of the old economy. The illusion of infinite software expansion that dominated capital markets for the past decade is over. Real-world algorithms demand massive amounts of cooling water to chill giant data centers and baseload power that does not falter for a single second. The current phenomenon, where Big Tech capital—driven by the fear of state-controlled public grid collapse—directly purchases nuclear power plants and pours astronomical capital into the SMR (Small Modular Reactor) value chain, marks the end of the ideological ESG narrative. It is the definitive beginning of a structural fracture where energy security dictates the survival and valuation of intelligence capital.

EXECUTIVE SUMMARY

As of [April 2026], the global macroeconomy is completely recalibrating valuations between real assets and tech companies, reflecting the 'baseload power shock' triggered by the explosive surge in AI data center power demand. The massive power consumption sparked by intelligence capital's computations proves the physical intermittency limits of renewable energies like solar and wind, systemically forcing infrastructure investments centered on SMRs and uranium. The phenomenon of U.S. hyperscalers making direct deals for nuclear power and building proprietary grids represents a macroeconomic inflection point where private capital privatizes national infrastructure sovereignty. This entrenches a structural deficit across the global uranium value chain, triggering a commodity supercycle, while simultaneously permanently crowding out innovative companies that failed to procure power.

01. Intelligence Capital's Power Predation and the Collapse of the Renewable Energy Illusion

└ The Massive Energy Intensity of Generative AI and the Overload Tipping Point of National Grids

As the parameters of ultra-large AI models expand exponentially, the energy consumption rate required for computation and cooling draws an exponential upward curve that transcends traditional Moore's Law. According to the [April 2026] global power grid load data released by the International Energy Agency (IEA), the processing cost of a single generative AI prompt consumes on average over 12.5 times more power than a traditional Google search, completely exhausting the transmission grid capacity in data center-dense regions. As the pace of national infrastructure expansion fails to keep up with the capital expenditure (CapEx) velocity of intelligence capital, extreme bottlenecks are occurring in key hubs like Virginia and Texas, where grid connection queues are pushed back by years. This translates directly into a fatal business risk of delayed AI model training for hyperscalers.

└ The Curse of Intermittency and the Forced Choice of Zero-Carbon Baseload Power

In the past, Big Tech companies drove the RE100 narrative, claiming to use 100% green energy through Power Purchase Agreements (PPAs) with solar and wind farms, but this system has completely gone bankrupt, colliding with physical contradictions. The fatal intermittency of renewable energy—where generation stops when the sun sets or the wind dies—is fundamentally incompatible with the fault-tolerant systems of AI data centers, which must unconditionally maintain a 100% utilization rate 24/7/365. Because economic viability fails even when deploying astronomically expensive Energy Storage Systems (ESS), the shift of the capital allocation axis toward nuclear energy—the only alternative that can provide continuous power while evading carbon emission regulations—is not a choice from a macroeconomic perspective, but a forced reversion for survival.

02. System Architecture Restructuring: Big Tech's Grid Privatization and the SMR Value Chain

└ Private Capital's Direct Purchase of Nuclear Plants and the Exodus from Public Infrastructure Systems

The most notable structural change in the [April 2026] macro system is the phenomenon of the 'Privatization of the Grid', where massive intelligence capital entities like Microsoft and Amazon break away from the existing state-led public grid ecosystem to sign direct power supply contracts with nuclear power plants. Constellation Energy's contract to restart the Three Mile Island nuclear plant or Amazon Web Services' (AWS) acquisition of the Talen Energy data center campus are not mere investments, but the result of a cold calculation that they cannot entrust the company's future computational power to slow, bureaucratized national transmission infrastructure. This signifies a fundamental restructuring of the system architecture, where private tech companies secure independent energy sovereignty to build a massive 'Intelligence-Energy Empire' outside of state control.

└ Capital Concentration in SMRs (Small Modular Reactors) to Bypass Large Nuclear Risks

The construction of conventional Large Nuclear plants is accompanied by massive initial capital, lengthy construction periods of 10 to 15 years, and astronomical cost overrun risks, making it incompatible with the financial structures of tech companies that must defend quarterly earnings. To overcome this, massive venture capital and Big Tech surplus cash are fiercely flowing into the SMR ecosystem, including TerraPower and NuScale Power. The SMR ecosystem, which will save the AI power grid, is an innovative system that deploys reactors in modular forms directly adjacent to data centers, eliminating transmission grid construction costs and reducing power loss to near zero. This is forcing an astronomical capital expenditure (CapEx) cycle upon the hardware infrastructure market, rivaling the second dot-com boom.

03. Geopolitical Collisions and Real Assets: The Uranium Chokepoint

└ The Structural Deficit of the Global Primary Uranium Production Network

Behind the nuclear renaissance lies a fatal macroeconomic detonator: the destruction of the uranium mining ecosystem, the core nuclear fuel. According to the [April 2026] supply-demand modeling by the World Nuclear Association (WNA) and major resource analytics agencies, due to more than a decade of underinvestment in mining infrastructure accumulated since the Fukushima disaster, primary mine production is recording a severe supply deficit of over 25% against annual global uranium demand. Given the geological characteristic that it takes over 10 years from mine discovery to actual extraction, a new supply chain capable of meeting the exploding demand in the short term does not physically exist. This operates as the trigger for a structural supercycle that forces global capital to flee from paper fiat into real assets like spot uranium and trust funds.

└ Geopolitical Weaponization and Bloc Formation of Enriched Uranium (HALEU) Supply Routes

The most fatal systemic flaw is that the global supply chain for High-Assay Low-Enriched Uranium (HALEU), essential for powering next-generation SMRs, is thoroughly fragmented. As of [April 2026], more than half of the commercial HALEU enrichment and refining infrastructure is controlled by geopolitical adversary Russia and Kazakhstan, which lies within its sphere of influence. As the Western bloc restricts Russian nuclear fuel imports for security reasons, SMR projects for AI data centers in the U.S. and Europe have hit a fatal chokepoint where they build reactors but cannot procure the actual fuel. This ultimately forces the West to inject massive duplicate capital to rebuild its own enrichment infrastructure, proving how the weaponization of resources exerts a permanent cost-push pressure on the macroeconomic system.

04. Macroeconomic and Valuation Distortions: The Revenge of Old Economy Hard Assets

└ Collapse of the ESG Narrative and the Re-rating of the Utility Grid Ecosystem

The explosive power demand of intelligence capital induces extreme valuation distortions within the capital market, completing the megatrend known as the 'Revenge of Real Assets'. Power utilities, ultra-high voltage transformer manufacturers, and uranium mining companies, which traded at P/E ratios under 10x for years crushed by ESG regulations and the fear of stranded assets, are being redefined as the core infrastructure bottlenecks of the AI revolution, experiencing historical multiple expansions that overwhelm tech companies. The cold macroeconomic law that intangible software cannot exist for a single byte without the physical hardware infrastructure that consumes massive energy is applying powerful re-rating pressure across the entire fiat money system.

└ The Crowding-Out Effect of Shadow Banking and the K-Shaped Polarization of Intelligence Capital

Big Tech's unlimited AI infrastructure CapEx sucks up the limited liquidity and scarce power grid resources within the macroeconomy like a black hole. An extreme winner-takes-all system is becoming entrenched, where only a few ultra-large hyperscalers capable of bearing the astronomical costs of securing the power grid maintain data sovereignty and top-tier AI models. Conversely, small and mid-sized AI venture capital unable to amass power by mobilizing massive financial and political power, and traditional manufacturing and commercial real estate sectors that failed to refinance in the shadow banking system due to high interest rates, are facing the cruel bill of deleveraging, permanently marginalized in the market as their credit creation mechanisms collapse.

05. Historical Comparative Analysis: The Forced Power Source Replacement Trajectory of the 19th Century Industrial Revolution

└ The Intermittency Limits of Hydropower and the Macro Shift to the Coal (Baseload Power) Ecosystem

The massive capital migration from renewable energy to nuclear power currently unfolding as of [April 2026] is systemically and perfectly identical to the macroeconomic trajectory during the First Industrial Revolution in the 19th century, when the power source of the textile industry shifted from hydropower (waterwheels) to steam engines and coal. Early Industrial Revolution capital utilized nature-dependent hydropower, but the intermittency of hydropower—restricted by lack of rain and limited locations—could not sustain the load of the exploding mass production systems. Ultimately, global capital poured astronomical amounts of money into building 'coal' mining infrastructure, which spouted baseload energy at any time regardless of weather and location, completing the backbone of railway networks and global capitalism. The current movement of intelligence capital, realizing the physical limitations of solar power and fiercely entering the uranium ecosystem, is an exact repetition of this massive historical macro cycle.

06. Variables and Limitations of the System Fracture Scenario

└ [Variable 1: Self-Healing & Exceptions] An Optimization Leap in AI Model Computational Algorithms

The most powerful variable that could delay or alter the trajectory of the nuclear-centric expansion scenario is intelligence capital's own 'Optimization' capability for computation. If the open-source camp or core Big Tech innovate neural network architectures to successfully commercialize lightweight models (sLLM) that compress power consumption to a tenth of current levels and inference-dedicated low-power NPU (Neural Processing Unit) chips early on, the extreme dependency on uranium and large-scale power grid infrastructure could be alleviated. This would act as a self-healing mechanism, flattening the exponentially soaring power demand curve and short-termly undermining the valuation premium of infrastructure-related assets.

└ [Variable 2: Resilience & Counter-Scenario Possibility] Licensing Regulations and Long-term Capital Lock-in

Despite the revenge of real assets, the most massive barrier facing the nuclear renaissance is not technology, but political vetocracy and regulatory systems. Due to the rigid environmental assessments by major countries, including the U.S. Nuclear Regulatory Commission (NRC), and severe NIMBY phenomena in local communities, there is a strong possibility that SMR and transmission grid licensing will be continuously suspended. If a long-term capital lock-in situation occurs—where Big Tech companies, despite injecting massive capital, fail to produce a single MW of power for years due to regulatory barriers, freezing their funds—the SMR expectations spread across the market will collapse, becoming the detonator for a counter-scenario where AI investment sentiment itself rapidly deleverages.

Macro Scenario: Probabilistic Future Trajectories

└ Scenario A (Base Case): Gradual Integration of the SMR Value Chain and Trending Upward Uranium Prices

First-generation commercial SMR projects backed by Big Tech capital clear regulatory hurdles and sequentially begin operations between [2028-2030]. While the life extension of existing large aging nuclear plants primarily defends against grid collapse, Western geopolitical subsidies are heavily injected into rebuilding the HALEU enrichment and uranium mining supply chains. Reflecting the structural supply deficit, spot uranium prices consistently raise their floor and trend upward, and the independent 'Microgrid' ecosystem—building data centers and reactors as a package—settles in, further consolidating the power of specific hyperscalers.

└ Scenario B (Structural Shift Case): Power Grid Blackouts Due to Regulatory Paralysis and Valuation Plunge

Trigger: Expansion of power grid infrastructure and SMR licensing are blocked by extreme vetocracy from environmental groups and local politics, and simultaneous massive power grid blackouts occur in core North American data center dense areas due to extreme weather like heat waves.
Result: Regulatory authorities invoke power rationing, forcibly assigning private AI data centers to the lowest priority for power usage. Consequently, orders are canceled across the entire AI chip value chain, including Nvidia, which had been granted valuations predicated on infinite expansion, triggering a brutal, tech-centric stagflation shock and valuation meltdown in the global capital markets.

└ Scenario C (Tail Risk Case): Blockade of the Eurasian Nuclear Fuel Supply Chain and Geopolitical Capital War

Trigger: Countering Western geopolitical pressure, the Eurasian resource bloc, including Russia and Kazakhstan, completely halts primary exports of the nuclear fuel cycle (mining, conversion, enrichment) and fully weaponizes uranium transport routes.
Result: Western SMR construction projects and existing large nuclear plants face a catastrophe where they are physically stranded, unable to secure fuel. This induces panic buying in the global spot uranium market, skyrocketing prices by multiples, and completes an irreversible geopolitical fragmentation where China and Russia, equipped with their own complete nuclear fuel ecosystems, seize global intelligence capital hegemony using state-owned data centers as weapons.

Implications from an Investor's Perspective

└ Short-Term (1-2 years from the date of writing)

As of [April 2026], this is the phase where grid bottlenecks materialize into real earnings shocks. Exposure to simple AI software companies or small-to-mid-sized cloud operators unable to control their massive power costs must be entirely reduced. Instead, to fill the void of delayed nuclear infrastructure, short-term liquidity must be concentrated on the essential manufacturing corporate groups (Picks and Shovels) related to global ultra-high voltage transformers, power cables, and power plant cooling systems, focusing on the aging grid replacement cycle where capital is immediately pouring in.

└ Medium-Term (3-5 years from the date of writing)

As Big Tech's astronomical SMR investments enter the actual licensing and construction phases, the sorting of true beneficiaries begins in earnest. Simple SMR design startups should be excluded from portfolios due to high regulatory friction risks. Medium-term capital should be parked in heavy industry companies monopolizing the forging production capacity essential for nuclear plant construction, global construction firms possessing Engineering, Procurement, and Construction (EPC) references, and monopolistic nuclear cycle companies sweeping up geopolitical subsidies to rebuild the West's HALEU nuclear fuel enrichment networks.

└ Portfolio Perspective

It is a period of systemic great transition where old economy hard assets grab intangible intelligence capital by the collar. The tech-centric 60/40 traditional portfolio loses its defensive capability in the face of grid fracture risks. To diversify the overall portfolio structure, building a powerful Barbell Strategy that hedges against inflation and supply chain blockade risks by incorporating physical uranium spot trust funds (SPUT) and geopolitical stateless assets like gold is more urgent than ever.

Conclusion

The expansion of massive intelligence capital (AI) is paradoxically pulling humanity forcefully into the swamp of the heaviest and most dangerous primary commodities: uranium and SMRs. The current macro phenomenon of Microsoft and Amazon spending tens of billions of dollars to buy nuclear plant sites is a market declaration of surrender, proving that the seemingly infinite data revolution ultimately falls under the dominion of the oldest and most finite physical infrastructure: 'power.' The nuclear fuel value chain, destroyed over the past decades while intoxicated by the superficial illusion of renewable energy, is now presenting the bill of structural supply deficits to the global capital markets. Investors must awaken from the seamless illusions created by software and perfectly reconstruct the survival foundations of their portfolios for this cold revenge drama of hard assets, where boiling cooling water and uncontrollable commodity prices tighten the leash around intelligence capital.

※ Disclaimer

This report does not solicit the purchase or sale of any specific assets, nor does it support or criticize any specific regime, government, or politician. It is a macroscopic system analysis article based on disclosed data and historical indicators. Not all market variables can be predicted, and the responsibility for all judgments and their resulting consequences lies with the reader. The author (Neutral Observer) does their utmost to ensure the reliability of the analysis but does not guarantee the perfect accuracy of the provided information.

Sources and References

[¹] International Energy Agency (IEA), Electricity 2026: AI Data Centers and the Collapse of Renewable Reliability (2026.04) — https://www.iea.org/reports/electricity-2026

[²] World Nuclear Association (WNA), Global Uranium Market Outlook and Structural Deficit Analysis (2026.03) — https://world-nuclear.org

[³] Center for Strategic and International Studies (CSIS), The Geopolitics of HALEU and the Advanced Nuclear Fuel Cycle (2026.02) — https://www.csis.org

[⁴] The Wall Street Journal, Big Tech's Nuclear Option: Amazon and Microsoft Bypass the Grid (2026.03.15) — https://www.wsj.com/business/energy-oil

[⁵] Goldman Sachs Global Investment Research, AI CapEx and the Revenge of the Old Economy: Grid and Uranium (2026.01) — https://www.goldmansachs.com/insights

[⁶] U.S. Nuclear Regulatory Commission (NRC), Advanced Small Modular Reactors (SMRs) Licensing Status and Bottlenecks (2026.03) — https://www.nrc.gov