DOSSIER: MACRO-STRUCTURAL ANALYSIS
I. THE AI CAPEX CYCLE: TEMPORAL HORIZON AND GROWTH DYNAMICS
The current AI infrastructure build-out is not a standard semiconductor cycle; it is a structural re-architecting of the global compute stack. We are currently in Phase 2 (Infrastructure Build-out) of a 4-phase transition.
| Phase | Description | Estimated Duration | Status |
|---|---|---|---|
| I: Proof of Concept | LLM breakthrough, initial H100 orders. | 2022 – 2023 | COMPLETED |
| II: Hyperscale Scaling | Data center densification, power grid integration. | 2024 – 2027 | ACTIVE |
| III: Edge Integration | On-device AI, inference-led hardware cycles. | 2026 – 2029 | PENDING |
| IV: Equilibrium | Replacement cycles and software-driven ROI. | 2030+ | PENDING |
Growth Rate Projections (CAGR):
- Hyperscaler Capex (MSFT, AMZN, GOOGL, META): Projected 32% CAGR through 2026.
- Energy Infrastructure (DC Power): Projected 18% CAGR as GPU thermal design power (TDP) moves from 700W to 1200W+.
- Inference Compute: Expected to flip from 20% of compute load to >70% by 2028, sustaining hardware demand even if "training" plateaus.
II. GEOPOLITICAL FRICTION MATRIX: CONTINUING CONFLICTS
ADAM's predictive models indicate a shift from "kinetic resolution" to "perpetual attrition" across three primary theaters.
The Russo-Ukrainian War (The War of Attrition): Likelihood of a frozen conflict is high. Expect long-term structural exclusion of Russian energy from G7 markets, cementing the "shadow fleet" economy.
Middle East Multi-Front (The Maritime Chokepoint): Conflict will likely persist via proxy. The primary threat is the normalization of trade route diversions (Cape of Good Hope vs. Suez), which adds 10-15% to global shipping costs permanently.
South China Sea/Taiwan (The Semiconductor Siege): Unlikely to go "kinetic" in the short term, but "economic blockade" simulation frequency has increased. This creates a "Just-in-Case" inventory tax on all tech firms.
III. COMMODITY IMPACTS: THE DUAL-DRIVE EFFECT
Commodities are being squeezed by two opposing forces: AI Electrification (Demand) and Geopolitical Fragmentation (Supply).
A. The "AI Metals" Complex
- Copper (Cu): The primary beneficiary. Data centers require 3x more copper than traditional offices.
Deficit Projection: 4-5 million tons by 2030. - Uranium (U3O8): The only viable path to carbon-neutral, 24/7 "Base Load" for hyperscalers.
Outlook: Structural supply deficit as Big Tech signs direct Power Purchase Agreements (PPAs) with nuclear operators. - Silver (Ag): Essential for high-frequency PCB manufacturing and solar PV expansion (powering the grid).
B. The Conflict/Energy Transmission
- Natural Gas (LNG): Becomes the "swing fuel" for AI power. US Henry Hub prices will decouple from historical norms as export capacity increases to bridge the European gap.
- Oil (Brent): A "Geopolitical Risk Floor" of $75-$80 is established. Any de-escalation is offset by OPEC+ discipline and US Strategic Petroleum Reserve (SPR) replenishment needs.
IV. RISK METRICS AND VALUATION IMPACTS
| Asset Class | Probability of Default (PD - 1yr) | Expected Value (EV) Shift | Narrative Shift |
|---|---|---|---|
| Mega-Cap Tech | 0.02% | +15% (Inference Growth) | From "Efficiency" to "Dominance" |
| Global Shipping | 4.50% | +20% (Rate Volatility) | From "Commodity" to "Strategic Asset" |
| EU Industrial | 8.00% | -10% (Energy Tax) | From "Core" to "Peripheral Risk" |
| Copper Miners | 1.20% | +35% (Supply Crunch) | From "Cyclical" to "Structural Growth" |
ADAM’S SYNTHESIS
The AI Capex cycle will persist through 2028-2030, but its composition will shift from chips to kilowatts. Investors should expect a transition where the bottlenecks move from Nvidia's H100 availability to Transformer-rated copper and Nuclear regulatory approvals.
Geopolitically, we are entering an era of "Fortress Economics." High growth in AI will coexist with fragmented supply chains, where commodity pricing is dictated by security of supply rather than spot-market efficiency.