Last week, NextEra Energy announced it is acquiring Dominion Energy for $67 billion. It is the largest utility merger in American history. The strategic rationale, stated plainly in the filing: AI-driven power demand.

AI power demand led the announcement.

This is worth sitting with for a minute. A power company looked at where electricity consumption is headed over the next decade, concluded that AI infrastructure is the dominant variable, and bet sixty-seven billion dollars on that conclusion. That would permanently restructure two of the largest energy companies in the country.

What the Infrastructure Commitments Reveal

Every week, there is a new model release, a new benchmark, a new agent demo. Those are capability signals. They tell you what AI can do today. They do not tell you whether the people building AI infrastructure believe this is permanent.

Capital allocation does.

NextEra is a regulated utility. Utilities plan across decades, using load forecasts, demand curves, and signed contracts. When a utility restructures at this scale, it is because the demand is already in the pipeline. The contracts commit customers to that demand.

And NextEra is not alone. SpaceX’s IPO prospectus, filed the same week, revealed that Anthropic is paying $1.25 billion per month for GPU compute through May 2029. That is $45 billion committed to a single infrastructure contract. Anthropic itself just projected $10.9 billion in revenue for Q2 2026 alone, up 130% from Q1. Its valuation is approaching $900 billion.

These are operational commitments based on current AI demand. The infrastructure layer is being built for permanence, and the numbers are no longer debatable.

The Part Most Leaders Are Missing

Here is where it connects to your business. The infrastructure being built right now is sized for organizations already using AI at scale. The power plants, the data centers, the compute contracts, the fiber builds. All of it is being provisioned based on current demand from current customers.

When that infrastructure comes online in 2028 and 2029, it will serve the organizations that were already in line. The ones that had workloads running, agents deployed, workflows automated. The ones that needed the capacity because they were already using it.

If your organization is still in the “exploring AI” phase by then, you will be competing for whatever is left after the early movers have locked in their capacity. This is how infrastructure advantages compound. The organizations that moved first get the best pricing, the best access, and the most reliable supply. Everyone else pays more for less.

Cloud computing provides a precedent for this pattern. The companies that migrated early got preferential pricing from AWS and Azure that late movers never received. But the AI version is moving faster and the capital involved is an order of magnitude larger.

What $67 Billion Means for the Gap

Belief or skepticism about AI is increasingly irrelevant. The infrastructure decisions have already been made. $67 billion in utility mergers. $45 billion in compute contracts. Hundreds of billions in data center construction globally. These commitments are not reversible. The power plants will be built. The capacity will exist. And it will be consumed by the organizations that planned for it.

The gap between AI-native organizations and everyone else includes an infrastructure gap. The early movers are locking in the physical resources that make AI possible at scale. That is a structural advantage that gets harder to close with every quarter that passes.

Six months from now, more capacity will be contracted. More infrastructure will be committed. More of the supply chain will be spoken for. The window is closing because the physical world is catching up to the digital one, and physical infrastructure takes years to build.

The Only Question That Matters

A $67 billion utility merger demands an operational response. The electrical grid is being restructured around AI workloads. The compute supply chain is being locked up by organizations already at scale. The investment horizon for this infrastructure is measured in decades.

The question is whether your organization will be positioned to access the infrastructure being built right now, or whether you will be standing in line behind the companies that moved while you were still evaluating.

That is an operations decision. And it has a deadline that no one is going to extend for you.