AI and energy financing are converging, with power becoming an equally critical constraint on data center expansion

Overview

The report focuses on the convergence of AI infrastructure financing and energy financing. The authors argue that the market had previously focused mainly on the scale of data center capital spending, but the more important constraints are now emerging in the upstream power system: generation capacity, transmission networks, power-grid equipment supply chains, and grid interconnection timelines are all growing more slowly than compute demand. Therefore, AI infrastructure construction will rely more on solutions that secure power in advance or obtain power faster, such as off-grid supply, energy storage, gas turbines, fuel cells, and the conversion of Bitcoin mining sites.

Core views

The core view is that the capital needs of AI infrastructure expansion are merging with the capital needs of the energy system, breaking the investment boundaries traditionally dominated by utilities. Insufficient power supply will make the buildout pace slower, more sequential, and more capital intensive, but it may also enhance the pricing power of companies with reliable compute capacity and power access. The report also points out that, at the current stage, enterprise demand for AI compute is relatively lacking in price elasticity; higher prices may not significantly suppress adoption and may instead push compute usage toward higher-value scenarios.

Analysis framework

The report uses a thematic macro framework, analyzing AI capital expenditure, equity and credit financing, power infrastructure, equipment supply chains, grid interconnection, labor, water resources, and public policy within a unified constraint system, while supplementing this with a weekly global macro events calendar covering rates, inflation, manufacturing, housing, and central bank policy.

Methodology notes

• Thematic macro analysisAI infrastructure financing constraint framework

  Assess data center capex together with power, grid interconnection, and energy asset financing

  The report argues that looking only at data center capex would underestimate the actual construction bottlenecks; generation, transmission, equipment lead times, interconnection queues, and financing channels must also be evaluated.

• Supply-demand and pricing power analysisCompute scarcity and the "compute merchants" framework

  When supply is constrained, participants with scalable and reliable capacity gain stronger pricing power

  When power and infrastructure constraints prevent compute supply from keeping up with demand, scarcity becomes a market characteristic and strengthens the bargaining power of companies that can provide stable capacity.

• Global macro trackingWeekly events calendar

  Use high-frequency macro events to validate growth, inflation, and policy paths

  The report lists forecasts for industrial production, housing, retail, inflation, employment, and central bank meetings across the United States, Europe, Japan, China, Australia, the United Kingdom, Brazil, and broader Asia.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• AI infrastructure and data centers

  Core demand side

  Strengths

  Enterprise AI demand is strong, current price elasticity is low, and high-value use cases continue to support compute consumption.

  Weaknesses

  Expansion is constrained by power availability, grid interconnection, equipment lead times, labor, water resources, and policy limits.

  Comparison

  The report emphasizes that this is no longer merely a data center capex theme, but a composite theme jointly financed with energy infrastructure.

  Risks

  Construction is slower than expected, capex payback periods lengthen, and local governments restrict new projects.

• Electric utilities and grid equipment

  Upstream bottleneck and key beneficiary segment

  Strengths

  AI data centers are driving rapidly growing incremental power demand, and scarcity in grids and equipment is increasing strategic value.

  Weaknesses

  The buildout cycles for generation, transmission, and transformer supply chains are clearly longer than the growth cycle of data center demand.

  Comparison

  The boundary between traditional utility investment and technology capex is being broken by AI construction demand.

  Risks

  Regulatory constraints, disputes over cost pass-through, worsening interconnection backlogs, and further deterioration in equipment lead times.

• Off-grid power, fuel cells, gas turbines, and energy storage

  Alternative solutions to shorten time-to-power

  Strengths

  They can provide faster and more customized power paths for data centers and are becoming direct contracting or financing targets for AI companies.

  Weaknesses

  There is uncertainty around solution costs, fuel sources, emissions constraints, and scalability.

  Comparison

  Compared with the traditional grid-connection model, these solutions place greater emphasis on speed and controllability.

  Risks

  Unstable project economics, tighter policy scrutiny, and execution risks in technology and supply chains.

• Scalable and reliable compute suppliers

  Potential pricing power winners

  Strengths

  In an environment of scarce compute supply, participants with stable capacity may become what the report calls "compute merchants."

  Weaknesses

  They must simultaneously manage power, capital, equipment, and customer demand commitments.

  Comparison

  Compared with pure data center operators, those with power access and financing integration capabilities have greater advantages.

  Risks

  If demand slows or supply is released in a concentrated way, pricing power may weaken.

• Equity and credit markets

  Financing channels for AI and energy capital expenditure

  Strengths

  Relevant transactions have already appeared in investment-grade and high-yield markets, reflecting that capital markets are beginning to price this converged infrastructure demand.

  Weaknesses

  Highly capital-intensive projects are sensitive to financing conditions and interest rates.

  Comparison

  Capital pools are shifting from relatively separate technology and utility investments toward a more integrated and interdependent financing model.

  Risks

  Wider credit spreads, uncertain project cash flows, and rising financing costs.

Key data

• Average lead time for power transformers128 weeksIndustrialSage says the current average lead time is 128 weeks, versus only 12-16 weeks before the pandemic.
• Lead time for generator step-up transformers144 weeksLead times for critical grid equipment have lengthened significantly, indicating that supply-chain constraints have become a construction bottleneck.
• U.S. interconnection backlogMore than 2x U.S. installed capacityBerkeley Lab says that by early 2025 the interconnection backlog had exceeded twice U.S. installed capacity, and the report believes it may be even longer now.
• U.S. electrician shortfallAbout 300,000 over the next decadeThe report says more than one-fifth of the existing electrician workforce is already age 55 or older, close to retirement.
• Share of data centers in high water-stress areas43%S&P analysis shows that 43% of global data centers are located in high water-stress regions, raising questions about sustainable expansion.
• U.S. state-level discussions of data center construction moratoriums14 state legislaturesThe report says 14 state legislatures are considering some form of data center moratorium measures.
• U.S. June ISM manufacturing PMI tracking estimate53.8The initial tracking estimate provided in the report's weekly macro observations.
• Forecast for the Bank of Japan's June meetingHike to 1.0%The report expects the Bank of Japan to raise rates to 1.0% at its June 15-16 meeting.
• UK CPI forecastHeadline 3.0%Y, core about 2.7%YThe report expects services, especially airfares, to drive UK inflation higher in May.

Impact & implications

The investment implication is that the bottlenecks in the AI theme are expanding from chips and data center capex to power, grid interconnection, equipment, energy storage, and financing structures. Beneficiaries may include assets with available power, reliable compute capacity, off-grid power solutions, and grid equipment capabilities; risks include longer construction cycles, rising capital intensity, increasing local policy resistance, and financing market volatility. At the macro level, the convergence of energy and AI capital demand may alter capital allocation in credit and equity markets.

Risks

• Power, grid interconnection, and transformer supply-chain constraints cause AI infrastructure deployment to lag market expectations.
• Shortages of skilled labor such as electricians raise project costs and extend construction timelines.
• Water stress limits continued expansion of water-intensive data centers in key hubs.
• State- or local-level moratoriums, cost-allocation disputes, and approval resistance are increasing.
• Tighter financing conditions may amplify execution risks for highly capital-intensive projects.
• If price elasticity of enterprise AI demand rises, compute pricing and utilization may come under pressure.

What to watch

• Whether lead times for power transformers and generator step-up transformers continue to lengthen.
• Whether the U.S. interconnection backlog continues to remain more than twice installed capacity.
• Whether AI companies continue to directly acquire, contract for, or finance power assets.
• Transaction and delivery trends for off-grid power, energy storage, gas turbines, fuel cells, and Bitcoin mining site conversion projects.
• Legislative progress across U.S. states regarding data center moratoriums and cost pass-through.
• Whether enterprise AI applications become more sensitive to rising compute prices.
• The impact of weekly macro events such as the FOMC, Bank of Japan, Bank of England, Reserve Bank of Australia, and Central Bank of Brazil on rates and financing conditions.