The AI boom is no longer just a chip story. It has pushed the owners of America's power grid into the center of Wall Street's next land grab.
Six months ago, the loudest AI trade was still Nvidia, cloud spending and the race to lease enough GPUs. Now look at the other side of the wall socket. The Financial Times reported on June 29 that US power and utility deals reached $203.6 billion in the first five months of 2026, already more than 40% above the total for all of 2025. You don't get a number like that because investors suddenly discovered substations are interesting. You get it because AI has turned electricity into a scarce strategic asset.
The biggest proof is NextEra Energy's proposed purchase of Dominion Energy. The Guardian reported that NextEra announced the roughly $67 billion all-stock deal on May 18, with the combined company set to serve about 10 million utility customer accounts across Florida, Virginia, North Carolina and South Carolina if regulators approve it. The FT put the broader transaction value at about $112 billion, including debt. Either way, this is not a sleepy utility merger. It is a fight for the territory where AI companies want to plug in.
Dominion's value starts in Northern Virginia, the world's densest data center market. Loudoun County and the wider Ashburn corridor already carry the kind of server load that most states would struggle to absorb, and the next wave of AI campuses will need still more generation, transmission and political patience. When NextEra moves for Dominion, it is not only buying meters and poles. It is buying a gate on one of the most important power routes in the AI economy.
The company knows regulators will not wave this through because investors like the story. NextEra has offered $2.25 billion in customer bill credits tied to the merger, a concession that Axios reported is already drawing scrutiny in Virginia. That is the fight you should watch. A one-time credit looks helpful on a bill, but ratepayers will ask the harder question: who pays when data centers force more spending on transmission lines, substations and generation?
Private capital wants the grid too
AES tells the same story from another angle. In March, a consortium led by BlackRock's Global Infrastructure Partners and EQT agreed to buy AES in a transaction valued at about $33.7 billion including debt, according to reports cited by the Financial Times. AES is not being taken private because public markets hate electricity. It is being taken private because the capital plan ahead is heavy, slow and difficult to explain every quarter.
Frankly, that is the part many AI investors still underplay. You can order chips faster than you can build a transmission line. You can announce a data center before a community has agreed to the noise, the water use or the rate impact. Utilities live inside that delay. Private equity sees the delay as an asset, because patient capital can own the bottleneck while everybody else argues about model training costs.
Alphabet has already made the logic explicit. Business Insider reported in December 2025 that Google parent Alphabet agreed to buy Intersect Power for $4.75 billion in cash, with assumed debt also part of the deal. Intersect is not just another renewable developer. It has been working with Google on co-located data center and power projects, including a site in Haskell County, Texas, where generation and compute are planned together instead of treated as separate problems.
That is vertical integration with a hard hat on. Google can sign power purchase agreements all day, but owning more of the development path gives it a different kind of control: where the power is built, when it arrives and how closely it matches the data center load. If you run a company that depends on AI infrastructure, this is the lesson. The winning firms are not only buying compute. They are buying the conditions that let compute exist.
Also read: xAI deploys Grok 4.5 at Tesla and SpaceX as Elon Musk bets his own companies on a 1.5 trillion-parameter challenger • South Korea bets $651 billion on AI and chips to challenge the global semiconductor order • China's GLM-5.2 matches a banned US AI on cybersecurity tasks and there's no export order that can stop it
The public backlash is not a side issue. Electricity prices, water demand and local land fights are now part of the AI supply chain. The FT noted that US electricity prices are up 9% nationally, while data center opposition has already become a local political force in several fast-growing markets. A utility merger that promises cheaper bills on day one can still leave customers exposed if the next ten years of grid spending are built around hyperscale demand.
That is why this M&A wave is bigger than dealmaking. It tells you where the AI economy is getting real. The early winners sold chips and cloud capacity. The next winners may be the companies that own the power plants, wires, permits and customer territories that make those chips useful. AI still looks digital from your screen. On Wall Street, it now looks like a utility map.
