AI Demand Is Only the Beginning. Infrastructure Is What Makes It Possible.

An AI infrastructure announcement can fit into a headline: a new campus, a multibillion-dollar investment, hundreds of megawatts of planned capacity.

Delivering that capacity takes considerably more.

Before a facility can support AI workloads, developers must establish a credible path through power delivery, network connectivity, site development, cooling, approvals, financing, construction, and operational testing. Each commitment depends on other commitments holding up.

AI demand creates the reason to build. Infrastructure coordination determines whether a facility can actually operate.

That relationship is central to Infrastructure of Intelligence™ (IOI), a research initiative examining the physical systems behind the Intelligence Economy and the companies, capital, and coordination required to build them.

What must happen between AI demand and an operating data center?

AI demand must become a defined computing requirement, a viable site, a deliverable power and connectivity plan, an approved facility design, committed financing, installed equipment, and tested operations. These activities overlap, and progress depends on coordinating their requirements and schedules.

For a new facility, the development process can be understood through seven practical steps:

1. Define the demand. Establish the customer, workload, computing requirements, and deployment schedule.

2. Validate the site. Determine whether the location can support the intended facility and its infrastructure.

3. Secure service pathways. Establish how electricity, connectivity, cooling resources, and any required fuel will reach the project.

4. Align design and approvals. Confirm that the technical design fits the site and applicable approval requirements.

5. Commit capital and procurement. Establish how construction and equipment will be funded and delivered.

6. Build and integrate. Install the facility’s systems and verify that they work together.

7. Commission and operate. Confirm that the completed facility can support its intended load and operating requirements.

This is an analytical outline, not a universal permitting sequence. A retrofit, leased deployment, and new campus will follow different paths.

The common requirement is that the physical systems must be ready together.

Why growing AI demand does not automatically create usable capacity

The scale of demand helps explain the pressure to build.

In its 2025 Energy and AI report, the International Energy Agency estimated that global data center electricity consumption would rise from approximately 415 terawatt-hours in 2024 to around 945 terawatt-hours in 2030 in its base case. AI was identified as the largest driver of that growth. These figures cover all data centers, including facilities supporting other digital services. Source: IEA, Energy and AI.

In the United States, Berkeley Lab’s 2024 analysis estimated that data centers consumed about 4.4% of national electricity in 2023, with a projected range of 6.7%–12% by 2028. That range reflects uncertainty about future growth. Source: U.S. Department of Energy, summary of Berkeley Lab’s report.

These forecasts establish the size of the infrastructure challenge. They do not establish whether an individual project has customers, financing, or a feasible delivery schedule.

A market forecast is not a customer contract. A land acquisition is not confirmation of utility service. And announced megawatts need a definition: they may refer to eventual campus capacity, an initial development phase, total facility power, or power available specifically to IT equipment.

For anyone evaluating the AI infrastructure market, those distinctions matter.

1. Translate AI demand into a facility requirement

“Demand for AI” is too broad to design around.

A developer or operator needs to understand what the customer intends to run, how much capacity it needs, and when that capacity must become available.

Questions include:

• Is the deployment intended for model training, inference, or a mix of workloads?

• What processors, storage, and networking will it require?

• What power density and cooling configuration must the facility support?

• What latency, availability, and security requirements apply?

• Will capacity be occupied immediately or delivered in phases?

These answers shape the building, its supporting infrastructure, and its commercial model.

A customer seeking a small initial deployment with room to expand presents a different development problem from one requiring a large block of capacity on a fixed date.

The first step is turning demand into a specification that engineers, utilities, developers, and investors can evaluate.

2. Validate the site across the eight pillars

A site’s value depends on what can be delivered there.

IOI examines that question through eight foundational pillars.

Power

Can sufficient electricity reach the facility with the required reliability and schedule?

The assessment must consider the service pathway, infrastructure upgrades, equipment, and conditions behind the proposed delivery date.

Connectivity

Can the site obtain the network capacity, latency, and route diversity its workloads require?

The assessment needs to establish how the facility will connect, what construction is necessary, and whether network routes meet its resilience requirements.

Land

Can the property accommodate the facility, supporting infrastructure, and planned expansion?

The site must work for the complete development, including access, utility infrastructure, cooling equipment, and other supporting uses.

Water

Can the cooling design function within local water availability, quality, and discharge constraints?

The answer depends on the complete cooling configuration and how heat will be rejected from the facility.

Workforce

Are qualified people available to design, build, commission, operate, and maintain the facility?

Workforce planning must address both the construction period and ongoing operations.

Policy

What approvals, utility rules, and public decisions shape development?

The project team needs to identify the applicable requirements, responsible authorities, and decisions that affect its schedule.

Capital

Can funding support the project’s costs, commitments, and delivery risks?

The financing plan must account for infrastructure, equipment, construction, and the period before operations generate revenue.

Sustainability

How will the project manage energy, water, emissions, and long-term resource performance?

These considerations belong in site selection and design because they influence how the facility will operate over time.

The eight pillars are IOI’s analytical framework. Their value is in making dependencies visible.

For example, a cooling decision can change electrical demand. That can affect utility requirements, equipment sizing, operating costs, and financing assumptions.

A site assessment therefore needs to examine how the systems fit together, as well as whether each is individually available.

3. Establish a credible path to power

A nearby transmission line is a starting point for investigation. It does not, by itself, establish that a facility can receive electricity.

The project team must determine available service capacity, required studies and upgrades, commercial terms, equipment needs, and the schedule for energization.

The IEA’s 2025 assessment estimated that roughly 20% of planned data center projects could face delays if grid risks were not addressed. It also reported that new transmission lines can take four to eight years to build in advanced economies. These findings describe system-level risks, not a delivery estimate for every project.

A practical power assessment should establish:

• The capacity available for the first operating phase.

• The pathway and conditions for future expansion.

• The infrastructure that must be built or upgraded.

• Responsibility for costs, construction, and maintenance.

• The remaining dependencies behind the proposed service date.

Onsite generation and storage can introduce additional options. They also introduce their own engineering, fuel, equipment, operating, and approval requirements.

The useful question is: What evidence supports the power delivery date?

4. Design connectivity and cooling around the workload

Electricity enables computation. Networks move data, and cooling keeps equipment within operating conditions.

Both must match the deployment.

For connectivity, a diligence review should establish the carriers that can serve the property, the construction required to connect it, available capacity, and the physical routes those services will use. Where resilience requires separate routes, the review should verify that separation rather than assume different providers use different infrastructure.

Cooling requires the same attention to the complete system.

“Liquid cooling” describes part of a thermal design. Evaluating water and energy requirements also means understanding how heat leaves the facility.

Microsoft illustrated this relationship in a December 2024 description of its next-generation data center design. The company described a closed-loop system designed to eliminate water evaporation for cooling, while acknowledging an energy trade-off from replacing evaporative cooling with mechanical cooling. Source: Microsoft, Sustainable by design: Next-generation datacenters consume zero water for cooling.

That example demonstrates why water, power, and sustainability decisions need to be evaluated together. A cooling technology label alone cannot establish a facility’s total resource requirements.

5. Align approvals, financing, equipment, and people

Once a project has a technically credible concept, it still needs a coordinated delivery plan.

The applicable approvals depend on the location and design. The project team must identify which decisions govern site development, utility infrastructure, construction, cooling, and any onsite generation.

Financing must account for the obligations being accepted along the way: land costs, infrastructure contributions, equipment deposits, construction payments, and the period before operations generate revenue.

Equipment procurement and workforce planning must follow the same schedule.

IOI’s analytical view is that these should be treated as connected commitments. Ordering equipment before a design is sufficiently settled can create rework. Waiting too long can put delivery dates at risk. Advancing construction without a credible service pathway can leave a completed building waiting for essential infrastructure.

A useful project plan makes four things explicit for every critical dependency:

What must happen, who owns it, what confirms completion, and what happens if it slips.

6. Build, integrate, and test the facility

Construction completion is a milestone. Operational readiness requires additional evidence.

Electrical distribution, backup systems, cooling, controls, networking, and safety systems must work together under the conditions the facility is intended to support.

Commissioning should verify performance against the project’s design and operating requirements. Depending on the facility, this can include testing equipment under load, checking control sequences, simulating power interruptions, and confirming how systems respond to failures.

The operating team also needs procedures, training, maintenance arrangements, monitoring, and clear responsibility for responding to incidents.

For phased campuses, readiness should be assessed by phase. An operating first building does not establish that the entire announced campus capacity is available.

Capacity becomes useful when the commissioned facility and installed IT systems can support the customer’s intended workloads.

Where infrastructure dependencies create partnership opportunities

The same dependencies that complicate development can reveal where companies need to work together.

Consider an illustrative project evaluating higher-density computing.

Supporting a new rack configuration

An IT supplier, cooling provider, and facility engineer may need to work together to validate power and thermal requirements. Their shared task is to establish whether the facility can support the proposed equipment and what modifications are required.

Delivering the first phase of electricity

A utility, developer, and electrical contractor may need to align service infrastructure and construction schedules. Their coordination helps connect the customer’s deployment date to the work required to energize the facility.

Establishing resilient connectivity

Network carriers, a fiber contractor, and the operator may need to confirm routes, capacity, and delivery responsibilities. The project needs evidence that the proposed network services meet its operational requirements.

Preparing for operations

The commissioning team, equipment suppliers, and operator may need to coordinate system testing and workforce training. Their shared work helps establish that the facility and its operating team are ready for service.

These are potential collaboration structures, not claims about existing partnerships.

The commercial opportunity becomes clearer when a company can identify a specific dependency it helps resolve.

For IOI, that is a central question in ecosystem strategy: Who needs to work together to move a project from stated demand to operating capability?

What leaders should ask before treating a project as ready

An announcement becomes more informative when it is accompanied by evidence.

What does the announced capacity measure?

Identify the development phase, power definition, and expected availability. Establish whether the figure describes the first operating phase or a longer-term campus plan.

What supports the demand?

Distinguish market interest from specific customer commitments. Understand who needs the capacity, what they intend to deploy, and when.

What supports the energization date?

Identify outstanding studies, infrastructure, equipment, and agreements. Determine which conditions must be satisfied before service can begin.

Are connectivity and cooling requirements validated?

Examine the actual service and design pathways. Confirm that they support the planned workloads and operating conditions.

Which dependencies remain unresolved?

Assign an owner and a completion milestone to each. Make clear how an unresolved item could affect the broader schedule.

What establishes operational acceptance?

Define the tests and conditions required before customer deployment. Establish who confirms that the facility is ready.

These questions help separate a promising concept from a project with a credible delivery path.

Frequently Asked Questions

What is AI infrastructure?

AI infrastructure includes the computing hardware, networks, facilities, and supporting systems required to train and run artificial intelligence. IOI examines its physical foundation through power, connectivity, land, water, workforce, policy, capital, and sustainability.

Why can an AI data center be delayed when demand is strong?

Strong demand does not resolve project dependencies. A facility can still be delayed by power delivery, site approvals, equipment availability, network construction, financing, workforce constraints, or incomplete testing.

Is an energized data center ready to run AI?

Energization confirms that electrical service has reached a particular stage. Readiness for AI workloads also depends on cooling, networking, installed IT equipment, commissioning, and operating capability.

What is the difference between announced and operating capacity?

Announced capacity describes a stated development plan. Operating capacity describes infrastructure that has been delivered and placed into service. Comparisons should use the same phase and capacity definition.

Understanding what must come together

The AI buildout requires companies that can design equipment, supply energy, connect sites, finance construction, and operate complex facilities.

It also requires those contributions to arrive in the right sequence, at the right location, with compatible requirements.

That is the work between demand and delivery.

Infrastructure of Intelligence™ examines how these systems depend on one another, where constraints emerge, and how those dependencies create opportunities for investment, development, and partnership.

Workforce

Policy

Capital

Sustainability