To AI and beyond

Executive Summary

The fundamental limit on global economic expansion is no longer algorithmic ingenuity or software distribution; it is thermodynamics, critical materials, and industrial infrastructure. As artificial intelligence advances into autonomous, multi-agent workflows, the foundational layer of economic value rotates from digital abstraction toward physical reality. Premium capital returns over the next decade will flow to the entities controlling the critical physical chokepoints of intelligence production. Translated into actionable public markets, this requires concentrating capital strictly into the North American industrial operators, metallurgists, and infrastructure integrators building the physical containers of compute.

1. The Thermodynamics of Intelligence

AI at scale is fundamentally a physics problem. Training and running frontier foundation models requires continuous, baseload power that municipal grids simply cannot support, fundamentally altering commercial real estate and infrastructure design.

• Behind-the-Meter Microgrids: Hyperscalers are increasingly operating as shadow utilities. Driven by multi-year transmission interconnection delays in Tier 1 markets, tech giants are investing heavily into off-grid, onsite power generation.
• The Nuclear SMR Renaissance: Small Modular Reactors represent the ultimate infrastructure solution for hyperscalers, offering co-located, carbon-free baseload power that completely bypasses municipal utility constraints.
• The End of Air Cooling: The mandate to dissipate extreme heat from next-generation silicon is driving a massive capital rotation into direct-to-chip liquid cooling and advanced thermal management, rendering legacy data centers obsolete.

2. Geopolitics, Supply Chains, and Sovereign Compute

Compute capacity has become a matter of state survival. The era of a borderless digital internet is permanently closed, replaced by a hyper-localized, heavily audited infrastructure paradigm.

• Sovereign AI Mandates: Nations are aggressively deploying capital to build localized, heavily secured compute stacks to maintain jurisdiction over their cultural data and national security, deliberately fragmenting the global hardware ecosystem.
• The Weaponization of Silicon: Export controls have transformed advanced semiconductor supply chains into geopolitical battlegrounds. Access to advanced lithography and high-bandwidth memory requires strict compliance and location-verification auditing.
• The New Resource Nationalism: Nations controlling critical minerals possess profound diplomatic leverage, fundamentally shifting geopolitical power toward resource-rich geographies.

3. The Un-Bypassable "Picks and Shovels" (Public Equity Proxies)

The true economic value of the AI infrastructure boom will be captured far upstream. To capture this in highly liquid, US-listed public markets, capital must rotate into the specific fuel fabricators, alloy producers, and heavy midstream integrators that un-block the SMR and data center pipelines.

• The Fuel Chokepoint & Commercial Nuclear Utilities: Next-generation reactors require High-Assay Low-Enriched Uranium (HALEU) and specialized fuel fabrication. Centrus Energy (LEU) stands as the vanguard of the US HALEU supply chain, while Cameco Corp (CCJ) captures baseline raw material contracts. Constellation Energy (CEG) serves as the direct proxy for "behind-the-meter" commercial nuclear baseload power agreements.
• Non-Fuel Critical Minerals & Superalloys: The specialized demands of exotic reactor coolants require materials like nuclear-grade zirconium and hafnium. With pure-play miners unavailable, capital flows to specialty metallurgists like ATI Inc. (ATI) and Carpenter Technology (CRS), which refine the exact corrosion-resistant superalloys required for extreme-environment energy infrastructure.
• Heavy Forging and Midstream Manufacturing: Factory-built modular reactors require flawless steel forgings for pressure vessels. The domestic pipeline relies on heavy defense contractors and midstream integrators like BWX Technologies (BWXT), Curtiss-Wright (CW), and GE Vernova (GEV) to provide the heavy-forging licenses, coolant pumps, and modular sub-assemblies.

4. The Hourglass Economy and the Collapse of Cognitive Arbitrage

The widespread deployment of autonomous agentic workflows is hollowing out the middle of the knowledge economy, forcing a global labor market rotation toward the extremes of strategic orchestration and physical reality.

• The BPO Disruption: The traditional "cognitive labor arbitrage" model—where developing nations exported human administrative bandwidth—is structurally obsolete. Economies reliant on cheap IT outsourcing must urgently pivot to specialized Global Capability Centers (GCCs), hardware assembly, or critical mineral refining.
• The Physical Trade Premium: Expanding the electrical grid, constructing nuclear campuses, and maintaining embodied robotics require specialized, highly skilled tradespeople. High-voltage electricians, precision machinists, and heavy infrastructure builders are commanding massive, sustained wage premiums.

5. Combating the Behavioral Sink

If basic cognitive and administrative survival is automated, human societies face the risk of a "behavioral sink"—stagnation in a frictionless, hyper-automated environment.

• The Authenticity Economy: To counter algorithmic atomization, societies will place an immense economic premium on the "hyper-real."
• The Relational and Liability Premium: Industries immune to automation—such as extreme physical endurance, artisanal craftsmanship, live performance, and complex relational care—will thrive. The assumption of human legal and moral liability will become a highly compensated, premium product.

6. Black Swans and Systemic Risks

This megawatt-and-hardware thesis must be hedged against non-linear disruptions that could shatter the current capital allocation trajectory. The most significant emergent threat is a demand-side failure, alongside technological, social, and geopolitical risks.

• The Automation Demand Death Spiral (The AI Layoff Trap): If AI displaces human workers faster than the economy can reabsorb them, it risks eroding the very consumer demand firms depend on. In a competitive economy, each firm captures the full cost savings from automation but bears only a fraction of the demand loss it creates in the broader market, as the rest falls on rivals. This demand externality traps rational firms in an automation arms race, displacing workers beyond what is collectively optimal. This macroeconomic aggregate demand failure fractures the infrastructure supercycle through three distinct vectors:The CapEx Transmission Shock: Hyperscaler revenue models are fundamentally downstream of broad consumer and enterprise cash flows. As rational firms aggressively automate, the aggregate collapse in consumer purchasing power shrinks B2B and B2C digital commerce and advertising revenues. Faced with stalling end-markets, cloud providers will abruptly curtail their multi-year capital expenditure run-rates, freezing the order books for direct-to-chip cooling, grid equipment, and heavy metallurgy.The PPA Bankability Crisis: Capital-intensive infrastructure, particularly behind-the-meter nuclear facilities and Small Modular Reactors (SMRs), requires 15-to-20-year Power Purchase Agreements (PPAs) underwritten by the flawless credit ratings of major tech giants. A demand-induced revenue shock threatens these corporate credit profiles, rendering these long-term baseload power projects un-bankable and stranding midstream integrators with zeroed-out pipelines.The Pigouvian Regulatory Threat: To halt a macroeconomic death spiral, regulators may intervene by implementing a Pigouvian automation tax. This regulatory friction, aimed at addressing the competitive incentives that drive rapid labor displacement, would artificially compress the ROI of enterprise AI adoption. This would drastically reduce the Total Addressable Market (TAM) for frontier models, rippling upstream to permanently cap compute hardware demand.
• Compute Deflation: Breakthroughs in sub-1-bit architectures, optical computing, or biological "organoid intelligence" could radically decouple AI from extreme energy consumption. Such an event would instantly deflate the necessity for hyper-scaled power generation and advanced cooling solutions, decoupling the software boom from the physical materials sector.
• Infrastructure Revolts: Mounting public anger over grid constraints, extreme water usage for data center cooling, and surging residential electricity bills could lead to severe municipal zoning restrictions. Localized political resistance has the potential to freeze the physical infrastructure supercycle before it is fully built.
• Physical Supply Chain Shocks: The hardware pipeline remains structurally fragile. A geopolitical blockade in the Taiwan Strait severing access to advanced logic chips, or targeted physical sabotage of highly specialized, difficult-to-replace high-voltage transformers, could halt the global AI rollout for years.

7. Macroeconomic Capital Rotation (Long-Only Construct)

Because systemic automation and physical chokepoints represent a one-way structural shift, capital allocation must concentrate strictly on a "Long/Overweight" portfolio composed of the physical containers of intelligence. The portfolio is structured across four primary pillars:

1. Sovereign Power Generation & Nuclear Infrastructure: Capturing the carbon-free, baseload megawatts through Constellation Energy (CEG), Vistra Corp (VST), and GE Vernova (GEV).
2. Thermal Management & Grid Equipment: Capturing extreme rack heat dissipation and high-voltage transmission step-ups through Vertiv (VRT), Eaton Corp (ETN), and Quanta Services (PWR).
3. Advanced Metallurgy & Fuel Fabricators: Capturing the physical constraints of reactor cores and advanced cooling loops through ATI Inc. (ATI), Carpenter Technology (CRS), Centrus Energy (LEU), and Cameco (CCJ).
4. Midstream Nuclear Integrators: Capturing the downstream assembly and component manufacturing of modular reactors through BWX Technologies (BWXT), Curtiss-Wright (CW), and Huntington Ingalls Industries (HII).