“The first phase established who could build. The next will establish who can earn.”
The AI Supercycle has crossed a financial threshold. With long-term rates above 5%, capital committed today must overcome a materially higher hurdle while AI assets face rapid depreciation and technological obsolescence.
At the same time, the stack is diverging. Model capability is proliferating, but packaging, power, cooling and grid access remain constrained. The result is a redistribution of pricing power toward the physical layers and a higher financial burden on companies funding capacity ahead of cash generation.
The emerging divide is Productive AI Capital vs. Trapped AI Capital. The fundamental question beneath the GVI is now clear: which corporations can translate structural AI leadership into durable cash returns before financing cost, depreciation and obsolescence absorb the advantage?
Top 10 Global Velocity Leaders
Structural driver × economic capture
The New Laws
The economics beneath the cycle
-
The AI cycle has created two financial paths. Compounding capital moves Capex → Revenue → FCF → ROIC → Reinvestment. Stranded capital runs Capex → Capacity → Depreciation → Financing Cost → Margin Pressure. The difference is not technological ambition — it is the distance between capital committed and cash generated. signal: That distance is now one of the most important corporate variables in the AI economy.
-
The physical stack — packaging, memory, networking, power, cooling, grid access — cannot expand at software speed. As intelligence becomes easier to reproduce, these harder-to-replicate layers gain relative strategic value. TSMC, Schneider Electric, Siemens Energy and Vertiv increasingly sit at this intersection. signal: The first-order AI trade was intelligence; the second-order trade is scarcity.
-
Power determines when infrastructure can begin producing revenue. Every delay adds financing cost, depreciation, obsolescence and deferred revenue — affecting project economics directly. signal: Time-to-power should be treated alongside capex, backlog and capacity as a core financial KPI.
-
The old assumption held that more backlog meant more visibility. The emerging hierarchy is Announced → Contracted → Financed → Powered → Cash-Generating. The farther backlog moves through the chain, the higher its economic quality. signal: Nominal backlog measures demand; economic backlog measures monetizability.
The Evolutionary Financial Dynamics
Five stages of the Supercycle
Sector Allocation Map
Economic necessity vs. capital intensity
Power & Grid
Electrical scarcity is the structural driver; the financial test is time-to-power plus cash conversion. Exposure: Schneider Electric · Siemens Energy.
Semiconductor Infrastructure
Manufacturing and packaging scarcity, tested on pricing and capital discipline. Exposure: TSMC.
Cooling / Data Centers
Compute density is the driver; deployment into cash the test. Exposure: Vertiv ecosystem.
Hyperscale Cloud
Infrastructure scale, tested on capex-to-FCF. Exposure: Microsoft · Amazon.
Enterprise Distribution
Workflow and customer ownership, tested on monetization. Exposure: Microsoft · Meta.
Agentic Security
Autonomous identities and permissions, tested on recurring revenue. Exposure: Palo Alto Networks.
Governed Deployment
Operational AI, tested on deployment-to-FCF. Exposure: Palantir.
Frontier Models
Technological leadership, tested on revenue versus fixed compute burden. Exposure: Anthropic.
Board-Level Implications
Five questions for the next dollar
Corporate Laggards — Bottom 10
Capital absorbed by repair
The shape of the next divide
- The AI cycle has moved from capital formation to return formation.
- Physical scarcity is concentrating pricing power in the layers hardest to reproduce.
- Higher-for-longer increases the cost of every delay between investment and cash generation.
- Backlog will increasingly be valued by economic quality, not nominal size.
- The next AI rotation occurs inside the stack: from technological creation toward bottlenecks, distribution, security and cash.