Artificial intelligence is hungry for electricity. As AI expands, the grid must keep up, and that need is shaping a new income opportunity. I focus on private infrastructure that supplies power to data centers. The goal is simple: stable, tax-advantaged yield, long-term contracts, and protection against rising prices.
Cars don’t work without gas, and AI is completely worthless without power.
The Core Idea
I target private facilities that generate or deliver electricity to data centers. These can be power plants, microgrids, substations, or dedicated transmission and distribution assets. In select structures, the income is tax-advantaged. That makes a 10% stated yield more powerful on an after-tax basis than many traditional options.
The demand is real. Today, AI already uses about 5% of U.S. electricity. The U.S. Department of Energy projects that it could reach 13% by 2030. This is not a small bump. It is a significant new source of load growth for the grid. More load means more infrastructure and longer-term contracts to support it.
I look for projects that sign multi-year power agreements with the largest technology companies. Think Meta, Microsoft, Amazon, and Google. These firms have strong balance sheets and huge ongoing computing needs. Long contracts with them can reduce the chance of default and bring consistency to cash flows.
View this post on Instagram
Why Yield Can Be Tax-Advantaged
The headline I study is a 10% yield with favorable tax treatment. Depending on the structure, distributions can be largely or entirely shielded from current income taxes. The results vary by investor, but the math is simple: untaxed income is worth more than taxed income at the same rate. For someone in a high bracket, a 10% untaxed yield can feel similar to roughly a 16% taxed yield. Actual outcomes depend on each person’s tax situation, of course.
Government policy supports the build-out. Energy projects linked to grid reliability and clean power may qualify for incentives and credits. Those incentives can help improve project economics. When the numbers work, it can lead to higher distributable cash flow to investors. Always confirm the tax treatment, the entity type, and the holding period with a qualified tax professional.
AI’s Electricity Appetite
Training and operating large AI models require a significant amount of power. As more companies adopt AI tools, the surge spills from research labs into daily business use. Data centers expand, and new campuses break ground. The result is a steady pipeline of power demand that does not turn off at night. These facilities run at high utilization to keep services online.
That continuity makes power contracts valuable. Data center operators place a premium on reliability, redundancy, and predictable costs. I look for agreements that price power in a way that shares risk fairly and adjusts over time. If a contract includes a fixed escalator or an inflation link, it can protect the income stream when prices rise.
A Real-World Example
Consider the scale of current development. Meta announced a roughly $10 billion data center project in Louisiana. Projects like this are not isolated. They are part of a nationwide buildout to meet compute and storage needs. Each new campus must secure reliable power. That is where private infrastructure comes in. If I can own or finance the facility that serves the data center, I can target income backed by a long-term agreement with a creditworthy tenant.
Long-term contracts also make project financing more predictable. When lenders see strong counterparties and committed offtake, financing terms improve. That can lower the cost of capital and increase the gap between project income and expenses. The end investor may receive that excess as yield.
How The Income Structure Works
Private infrastructure projects often use project-level companies. The project signs a power purchase agreement (PPA) or a similar contract with the tenant. The contract sets price, volume, and escalation terms. Revenues flow to the project company, which pays operating costs, services debt, and then distributes cash to investors.
Here is what I look for when I evaluate a deal:
- Yield: A target of around 10% with favorable tax treatment.
- Term: Contracts of 15 to 20 years or more.
- Credit: Counterparties like Meta, Microsoft, Amazon, and Google.
- Inflation Protection: Escalators tied to inflation or fixed step-ups.
- Alignment: Incentives that encourage all parties to maintain the plant’s online efficiency.
Electricity inflation has been running hot. A 7% rate can erode purchasing power—contracts with an inflation kicker help. If inflation runs at 4%, a 10% base yield could step to roughly 14%. Terms vary, so the details matter. The goal is to keep real income steady even when prices rise.
Why Tenant Quality Matters
Revenue risk is central. I want tenants who pay on time and honor long-term agreements. Mega-cap technology companies offer that profile. They are building for the long haul and need clean, reliable power to support AI growth. Their scale brings negotiating power but also a desire for stability. That combination can produce well-structured agreements.
High-quality tenants do not eliminate all risks, but they significantly reduce one major one: non-payment. That leaves me to focus on construction, operation, regulation, and market risks, which are easier to price and manage when the revenue side is strong.
Inflation Kicker And Real Return
Contracts often include escalators. Some step up at a fixed rate each year. Others are tied to inflation indices. In a period of rising power and equipment costs, that feature helps keep real income from slipping. It also supports the project’s ability to cover maintenance, upgrades, and debt service as costs move.
In practice, the inflation link provides a cushion. If inflation runs at 4%, a 10% yield can scale to about 14% under the right terms. This helps protect purchasing power. It is not perfect, but it is far better than a flat coupon when the cost of living rises.
Risk Factors I Watch
No yield is free of risk. I approach these projects with a clear checklist.
Construction and timeline risk: Projects can run late or go over budget. Strong engineering partners and fixed-price contracts help. Performance bonds and liquidated damages can add another layer of protection.
Operational risk: Plants must run safely and efficiently. I review operator experience, maintenance planning, and spare parts. Uptime targets and penalties in the contract focus everyone on reliability.
Regulatory and permitting: Energy projects need approvals that can take time. I prefer sites with permits in hand or clear pathways to secure them. Community relations also matter for long-term stability.
Commodity exposure: If the project burns fuel, I want to see clear hedges or pass-through pricing. If it is renewable, I examine resource data and curtailment risk. Weather-adjusted projections beat rosy assumptions.
Interest rates and refinancing: Higher rates squeeze margins. Fixed-rate debt and conservative leverage help. I model sensitivities to rate moves and stress-test coverage ratios.
Tenant concentration: One tenant can be both a strength and a single point of failure. I assess the tenant’s site plans, growth needs, and contract exit clauses. The stronger the need for the site, the better my odds over time.
Technology change: Efficiency gains can help tenants use less power per compute unit. But overall demand can grow faster than efficiency. I watch both factors and plan for multiple scenarios.
Why Private Infrastructure Fits My Playbook
I look for assets that address an obvious need, use simple contracts, and produce visible cash flow. Power for AI checks all three boxes. The need is clear. The contract is the backbone. The cash flow is evident when electrons flow and invoices are issued.
Another edge is alignment with policy. Policymakers want reliable power and more clean capacity. Incentives encourage investment, which can improve project returns. I do not rely on subsidies alone, but I like it when the policy tailwind supports the economics rather than defines them.
How I Think About Allocation
I do not treat private infrastructure as a replacement for core holdings. It is a compliment. The goal is to add a source of income that is not too tied to stocks or traditional bonds. Long-term contracts help lower volatility in cash flows. That can steady a portfolio when markets swing.
Position sizing depends on an investor’s liquidity needs, time horizon, and risk tolerance. These are private deals, so they tend to be less liquid than public markets. I assume a long holding period and size positions accordingly. Diversification across tenants, regions, and technologies helps manage risk.
What Could Go Right
If AI demand tracks the DOE’s view, the grid will need a surge of new capacity. Data centers will continue to seek dedicated power with strong reliability. That means more contracts and more projects. Over time, scale can bring cost advantages and operating know-how. Investors may see stable income and potential upside from rising demand and contract escalators.
What Could Go Wrong
The grid could hit bottlenecks in transmission. Permitting delays can stall projects. Fuel costs might spike without adequate hedging. Interest rates could rise and pinch margins at refinancing. Policy could change. Any of these would reduce returns. That is why the underwriting must be disciplined and the structure sound.
Pulling It Together
My thesis is straightforward: AI cannot grow without power, and power projects can provide investors with a steady income. A target 10% yield with favorable tax treatment, multi-decade contracts, investment-grade tenants, and inflation protection form a strong base. The opportunity is not about hype. It is about real assets tied to real demand.
The need is simple: more electrons for more compute. Private infrastructure supplies them, and long-term contracts can turn that into income.
Key Takeaways
- AI already uses about 5% of U.S. electricity; DOE sees up to 13% by 2030.
- Private power assets can target a 10% yield with favorable tax treatment.
- Long-term contracts with mega-cap tech can lower revenue risk.
- Inflation escalators help protect real income as costs rise.
- Risks include construction, regulation, commodity prices, rates, and concentration.
I expect demand for reliable data center power to keep climbing. The grid needs private capital, and policy supports new builds. That creates room for disciplined investors to seek high, tax-advantaged income. As always, verify the structure, review the contracts, and stress-test the numbers. Simple, durable cash flow wins.
