AI Data Center Investment: What Makes a Project Financeable

AI demand alone does not make a data centre financeable. How investors and lenders test power, customer contracts, the financial model, capital structure, governance and resiliency.

Paul Raftery3 September 2026, 11 min read

AI Data Center Investment is moving from a technology story to an infrastructure and capital-allocation challenge. Demand for AI compute is expanding rapidly, but institutional investors still need to determine whether an individual project has secured the power, customers, infrastructure, capital structure and management capability required to operate successfully.

The scale of the opportunity is significant.

The International Energy Agency reported in April 2026 that electricity demand from data centers increased by 17% in 2025. The IEA expects global data-center electricity consumption to double by 2030, while electricity use by AI-focused data centers could triple.

Read the IEA's 2026 analysis of data-center electricity demand https://www.iea.org/news/data-centre-electricity-use-surged-in-2025-even-with-tightening-bottlenecks-driving-a-scramble-for-solutions

JLL's 2026 Global Data Center Outlook projects nearly 100 GW of new data-center capacity between 2026 and 2030 and estimates that the infrastructure expansion could require up to US$3 trillion of investment, including real estate and IT fit-out capital.

Read the JLL 2026 Global Data Center Outlook https://www.jll.com/en-us/insights/market-outlook/data-center-outlook

Those figures demonstrate the scale of AI infrastructure investment.

They do not mean that every announced AI data center is financeable.

For institutional capital, the fundamental question remains:

Can the project convert demand for compute into an infrastructure asset with credible, resilient and testable cash flows?

AI Data Center Investment: What Institutional Capital Tests

Institutional capital looks beyond AI demand.

Professional investors and lenders generally need to understand how the complete investment case fits together:

Power → Site → Infrastructure → Customers → Revenue → Costs → Capital Structure → Risk → Return

Weakness in any part of that sequence can affect the rest.

An AI data center may have prospective customers but insufficient power.

It may have secured power but lack adequate interconnection certainty.

It may have a strong site and technical design but insufficiently contracted revenue.

Or the project may have all three but a Financial Model that does not adequately reflect construction delays, technology refresh, financing costs or downside cases.

Investor Readiness therefore requires more than demonstrating demand for AI.

It requires demonstrating that the project itself can be evaluated professionally.

Why Power Has Become Central to AI Data Center Investment

Power availability is one of the most important constraints shaping the data-center market.

The IEA reports that data-center electricity consumption increased rapidly in 2025 while projects encountered physical constraints involving grid connections, transformers, gas turbines, advanced chips and other equipment. Technology companies responded in part through renewable PPAs, nuclear agreements and interest in on-site generation.

Read the IEA's latest Energy and AI analysis https://www.iea.org/news/data-centre-electricity-use-surged-in-2025-even-with-tightening-bottlenecks-driving-a-scramble-for-solutions

JLL goes further in its 2026 outlook, identifying power availability, rather than simply location or cost, as a primary site-selection consideration in an environment where grid connection can involve multi-year waits.

Read JLL's 2026 Data Center Market Outlook https://www.jll.com/en-us/insights/market-outlook/data-center-outlook

What Investors Need to Understand About Power

Depending on the project, due diligence may examine:

Available power capacity
Grid connection
Connection timetable
Utility agreements
Power pricing
PPA structure
On-site generation
Backup generation
Transmission constraints
Curtailment
Battery storage
Fuel supply
Renewable-energy strategy
Regulatory approvals
Resiliency

The existence of power somewhere on a regional grid is not the same as having deliverable power available to a specific site when the data center requires it.

That distinction can materially affect construction schedules, capital requirements and revenue commencement.

Power Supply Is Also Becoming More Diverse

The IEA projects that electricity generation serving data centers will exceed 1,000 TWh by 2030 in its base case.

Renewables are expected to meet nearly half of the additional data-center electricity demand through 2030, while natural gas and coal together are projected to meet more than 40% of incremental demand over the same period. Nuclear becomes increasingly relevant toward the end of the decade and beyond.

Read the IEA analysis of energy supply for AI https://www.iea.org/reports/energy-and-ai/energy-supply-for-ai

For a project sponsor, this means power strategy should not be treated as a paragraph in the Information Memorandum.

It needs to be reflected directly in the Financial Model, construction schedule, permitting strategy and risk analysis.

Technology Matters, but Flexibility Matters Too

AI data-center technology is developing quickly.

That creates a challenge for long-lived infrastructure.

A facility designed around today's hardware needs to consider what happens when future compute requirements change.

JLL's 2026 outlook anticipates that AI could account for around half of data-center workloads by 2030 and highlights increasing rack densities and greater liquid-cooling requirements.

Read JLL's 2026 Global Data Center Outlook https://www.jll.com/en-us/insights/market-outlook/data-center-outlook

Uptime Institute's 2026 research similarly identifies high-density AI infrastructure, power constraints and resiliency as important issues for data-center operators.

Read Uptime Institute's Five Data Center Predictions for 2026

Technical Due Diligence May Examine

Facility design
Rack density
Cooling architecture
Power distribution
Redundancy
Network connectivity
Technology-refresh flexibility
Equipment supply chains
Construction sequencing
Commissioning
Resiliency
Operational capability

The investment question is not simply whether the facility can support today's hardware.

It is also whether the project's economics acknowledge that technology, cooling and power requirements can evolve during the investment period.

Customer Commitments and Revenue Visibility

High demand for AI compute does not automatically create bankable project revenue.

Institutional investors need to understand who will pay the project and on what contractual basis.

Potential counterparties might include:

Hyperscalers
AI model developers
Cloud platforms
Enterprises
Governments
Research institutions
Colocation customers
Neocloud providers

Investors May Test

Contract duration
Customer credit quality
Capacity commitments
Minimum payments
Renewal provisions
Termination rights
Pricing escalation
Customer concentration
Contracted versus uncontracted capacity

A letter of intent and a binding capacity agreement should not be modelled as if they provide the same revenue certainty.

Likewise, projected market demand should not be presented as contracted project revenue.

The Financial Model should distinguish clearly between:

Contracted revenue, supportable assumptions and potential upside.

The Financial Model Must Connect Power to Revenue

For an AI data center, the Financial Model needs to connect physical capacity with commercial and financing assumptions.

Depending on the project, important drivers may include:

MW available
MW contracted
IT load
Rack deployment
Utilisation
Pricing
Customer concentration
Power cost
Cooling
Construction cost
Equipment
Maintenance
Staffing
Network costs
Land
Tax
Financing
Technology refresh
Working capital

Base Case and Downside Cases

A professional Financial Model should not rely exclusively on one outcome.

Sensitivity analysis can test questions such as:

What if grid connection is delayed?
What if construction costs increase?
What if customer ramp-up is slower?
What if utilisation is lower?
What if power becomes more expensive?
What if a major customer does not renew?
What if technology-refresh expenditure occurs earlier?
What if debt is more expensive?
What if additional equity is required?

This testing helps management understand which assumptions have the greatest effect on the investment case before external due diligence begins.

Capital Structure: Debt, Equity and Infrastructure Risk

AI data centers can require substantial amounts of capital.

The correct structure depends on:

Development stage
Contracted revenue
Power certainty
Sponsor
Customer credit
Construction risk
Technology risk
Jurisdiction
Financing market
Asset ownership

Potential sources of capital may include:

Sponsor equity
Institutional equity
Infrastructure funds
Bank debt
Private credit
Project finance
Joint venture capital
Strategic capital

There is no credible universal sponsor-equity percentage or debt ratio that applies to every project.

The structure should emerge from the underlying economics and risk allocation.

What Capital Providers May Examine

Sponsor commitment
Equity funding
Debt service
Financing assumptions
Contingency
Liquidity
Completion support
Reserve accounts
Covenant headroom
Refinancing risk
Downside cases

The objective is not to maximise debt.

It is to construct a financing structure that remains credible under the project's realistic operating scenarios.

Governance and Execution Capability

AI infrastructure combines real estate, power, construction, technology, telecommunications and operations.

Governance therefore matters.

Professional investors may examine:

Ownership
Decision-making authority
Board structure
Management experience
Financial controls
Reporting
Construction oversight
Data-center operating experience
Cybersecurity governance
Risk management
Succession
Conflicts of interest

First-time sponsors are not automatically excluded from institutional funding.

But where direct operating experience is limited, investors may place greater weight on:

Experienced operating partners
EPC capability
Technical advisers
Independent verification
Conservative assumptions
Governance structures

The team does not need to claim it knows everything.

The investment case does need to demonstrate that the right capabilities exist around the project.

Resiliency Is an Investment Issue

Uptime Institute's Global Data Center Survey 2026 describes a sector continuing to expand while operating under increasing constraints.

The report highlights:

Limited power availability
Rising costs
Supply-chain constraints
Staffing shortages
Increasing high-density workloads
Continuing importance of resiliency

It also reports that more operators are seeing peak rack densities of 30 kW or above.

Read the Uptime Institute Global Data Center Survey 2026 https://intelligence.uptimeinstitute.com/resource/uptime-institute-global-data-center-survey-2026

Infrastructure investors therefore need to consider not only expected utilisation but the project's ability to operate reliably.

Resiliency may involve:

Power redundancy
Cooling redundancy
Network diversity
Backup generation
Maintenance
Spare-parts strategy
Staffing
Security
Disaster recovery
Operational procedures

Resiliency has a cost.

But failure also has a cost.

The Financial Model needs to reflect both.

AI Data Centers in 2026: Demand Meets Infrastructure Constraints

The scale of investment is significant.

JLL estimates that nearly 100 GW of new data-center capacity could be added between 2026 and 2030, potentially doubling global capacity. Its 2026 outlook estimates up to US$3 trillion of investment could be required across real estate and IT infrastructure.

AI could account for half of data-center workloads by 2030, according to JLL's outlook.

Read JLL's 2026 Global Data Center Outlook https://www.jll.com/en-us/insights/market-outlook/data-center-outlook

The IEA reports another side of that growth.

Data-center electricity demand increased 17% in 2025, and rising demand is encountering physical constraints in grids, permitting and equipment supply.

Read the IEA's April 2026 update https://www.iea.org/news/data-centre-electricity-use-surged-in-2025-even-with-tightening-bottlenecks-driving-a-scramble-for-solutions

This creates an important distinction:

Demand for AI infrastructure can be strong while individual projects remain difficult to execute.

Institutional capital therefore needs to evaluate both market demand and project-specific deliverability.

What Institutional Capital May Test

The questions investors ask should flow from the investment case.

Is Power Actually Available?

Strong evidence may include:

Executed agreements
Utility confirmation
Interconnection status
Technical studies
Pricing structure
Delivery timetable

Who Is Buying the Capacity?

Relevant evidence may include:

Executed customer contracts
Capacity commitments
Credit information
Minimum payments
Contract duration

Can the Project Be Built for the Modelled Cost?

Evidence may include:

Engineering
Contractor quotations
Procurement strategy
Construction schedule
Contingency
Independent cost review

Can It Operate Reliably?

Investors may examine:

Operations team
Maintenance
Redundancy
Cooling
Power systems
Network design
Staffing
Cybersecurity
Disaster recovery

Does the Financial Model Survive Downside Scenarios?

The model should demonstrate the effect of:

Delays
Cost increases
Lower utilisation
Higher power prices
Financing changes
Customer loss
Technology refresh

Is Governance Appropriate for Institutional Capital?

Investors may examine:

Ownership
Board oversight
Financial controls
Reporting
Risk management
Decision rights

Strong answers are generally specific and supported by documentation.

Narrative alone is not evidence.

Investor Readiness for AI Data Center Investment

An AI data center project should be able to connect the complete investment proposition:

Site → Power → Technology → Customers → Revenue → Costs → Capital → Risk → Return

The Financial Model should reconcile with:

Information Memorandum
Investor Pitch Deck
Teaser
Technical studies
Power documentation
Customer documentation
Data Room

If one document assumes 100 MW of capacity while another assumes 80 MW, investors will notice.

If the Financial Model treats prospective customers as contracted revenue, investors will notice.

If the construction schedule assumes power is available before the interconnection date, investors will notice.

Investor Readiness is about identifying and resolving these issues before professional capital evaluates them.

Projects RH prepares capital-intensive projects for professional investor review by establishing the facts, developing or testing the Financial Model and preparing consistent institutional documentation.

See how Projects RH works https://projectsrh.com/how-we-work.html

Capital raising and project finance are considered subsequently under separate mandates. Investor interest, financing terms and transaction completion are never guaranteed.

Frequently Asked Questions About AI Data Center Investment

What Makes an AI Data Center Attractive to Institutional Capital?

There is no single test.

Institutional investors generally need sufficient evidence to evaluate power availability, customers, revenue, construction, operations, management, governance, capital structure and downside risk.

Is There a Minimum Size for Institutional AI Data Center Investment?

There is no universal MW threshold.

The appropriate scale depends on investor mandate, transaction costs, location, customer commitments, power availability, economics and development stage.

A smaller project with stronger contractual support can present a different risk profile from a much larger speculative development.

How Important Is Power Supply?

Power supply is fundamental.

Both the IEA and JLL identify power availability and grid constraints as important factors shaping data-center development.

A project's power case needs to be evaluated alongside its customer, construction and financing assumptions.

Do AI Data Centers Need Long-Term Customer Contracts?

Contractual requirements vary by financing structure.

However, investors generally need to understand how much revenue is contracted, which counterparties support it, how long contracts last and what termination or renewal provisions apply.

Potentially.

A first-time sponsor may need to demonstrate the required capabilities through experienced management, operating partners, contractors, advisers and governance arrangements.

What Financial Model Sensitivities Should Be Tested?

Depending on the project:

Power cost
Grid connection
Construction cost
Construction delay
Customer ramp-up
Utilisation
Pricing
Financing cost
Customer concentration
Technology-refresh expenditure

How Does AI Change Traditional Data Center Investment?

AI can increase compute density, power requirements and cooling demands.

JLL expects AI to represent an increasing share of global data-center workloads, while Uptime Institute identifies high-density infrastructure and power constraints as major operating considerations.

Does Investor Readiness Guarantee Financing?

No.

Investor Readiness improves the quality, coherence and traceability of the information available for investment evaluation.

Capital providers make independent investment decisions and conduct their own due diligence.

Sources

International Energy Agency: Data Centre Electricity Use Surged in 2025

April 2026 analysis of AI, data-center electricity demand, power constraints and energy infrastructure. https://www.iea.org/news/data-centre-electricity-use-surged-in-2025-even-with-tightening-bottlenecks-driving-a-scramble-for-solutions

International Energy Agency: Energy Supply for AI

IEA projections for electricity supply supporting data centers through 2030 and 2035. https://www.iea.org/reports/energy-and-ai/energy-supply-for-ai

JLL: 2026 Global Data Center Market Outlook

Current outlook covering AI demand, power constraints, development capacity and investment requirements. https://www.jll.com/en-us/insights/market-outlook/data-center-outlook

Uptime Institute: Five Data Center Predictions for 2026

Analysis of power constraints, AI infrastructure, resiliency and data-center operations. https://uptimeinstitute.com/resources/research-and-reports/five-data-center-predictions-for-2026

Uptime Institute: Global Data Center Survey 2026

July 2026 research on costs, power availability, resiliency, rack density, staffing and AI-related workloads. https://intelligence.uptimeinstitute.com/resource/uptime-institute-global-data-center-survey-2026

Prepare Your AI Data Center Investment for Institutional Review

If your AI data center project is preparing for institutional equity, infrastructure debt, project finance or strategic capital, start with the evidence already available.

Already have a Financial Model, power documentation, customer agreements, technical studies, Information Memorandum or project materials?

Send your existing project documents to Projects RH https://projectsrh.com/intake.html

Want to understand how Projects RH prepares projects before capital raising?