Author | Hua Lin Wu Wang
Editor | Jing Yu
If someone had told me a few years ago that AI would ultimately reshape America's power grid landscape, you probably wouldn't have taken it too seriously. After all, we're talking about software, algorithms, model parameters — things that don't sound like they have much to do with power plants.
But on May 18, 2026, that perception was completely shattered.
NextEra Energy announced the acquisition of Dominion Energy for $67 billion, setting a record for the largest utility merger in U.S. history.
The number itself is staggering, but what's even more noteworthy is the underlying logic driving this deal. It's not traditional energy strategy, but the insatiable hunger for electricity from global AI data centers.
01 The Invisible 'Computing Blood Vessels'
To understand this merger, you first need to know about a place — Northern Virginia, specifically the Loudoun County area, known within the industry as 'Data Center Alley'.
This area hosts the world's densest concentration of data centers, where vast numbers of servers for AWS, Microsoft, Google, and Meta are located on seemingly ordinary land. It's estimated that about 70% of the world's internet traffic passes through here daily. And Dominion Energy is the primary electricity supplier for this region.
Dominion holds over 51 GW of contracted data center demand — what does 51 GW mean? Roughly equivalent to the installed capacity of about 50 large nuclear power plants, and this number is still growing. The load in Virginia's Dominion region is projected to grow by 121% by 2045.
This is the core reason NextEra is willing to pay $67 billion — not to buy a traditional power company, but to acquire the most scarce resource of the AI era: the 'power supply rights' adjacent to the heart of computing power.
The market has spent two years pricing AI chips; now it's starting to price the grid.
02 The Struggling Grid
Placing this merger within the timeline of the past year reveals it's not an isolated event, but the latest link in a chain reaction.
Rewind to 2025: IEA data had already sounded the alarm.
Global data center electricity demand surged by 17% in 2025, while overall global electricity demand grew by only 3%. Growth for AI-dedicated data centers left the broader market far behind. The IEA predicts that by 2030, global data center electricity consumption will double from 415 TWh in 2024 to about 945 TWh — most of that 530 TWh increase is attributable to AI training and inference workloads.
The combined capital expenditure of the five major tech giants in 2025 exceeded $400 billion, with a significant portion flowing into data center construction — and this number is expected to grow by another 75% in 2026.
The grid is starting to buckle under the pressure.
Just two days before this merger was announced, on May 16, a report from Monitoring Analytics revealed a disturbing reality: electricity prices in PJM Interconnection, America's largest power market, showed 'irreversible' sharp increases, up by 76%. PJM covers more than ten states including Virginia, Maryland, and Pennsylvania — precisely one of the most AI infrastructure-dense regions.
The report used the term 'irreversible' unusually. It's not talking about an adjustable price fluctuation, but a fundamental shift in the structure of electricity supply and demand.
Earlier, at the end of 2025, Northern Virginia experienced a real-world stress test of the grid. Voltage fluctuations caused 60 data centers to disconnect simultaneously, instantly creating a 1,500 MW power surplus — this sudden energy shock reminded everyone how fragile AI infrastructure is to grid stability and how demanding its power supply requirements are.
03 NextEra's Bet
NextEra is not an ordinary traditional utility. It is America's largest wind and solar power generator, with deep expertise in building and operating new energy infrastructure. Acquiring Dominion isn't just a simple scale expansion.
Combining NextEra's clean energy and storage capabilities with Dominion's market position in Data Center Alley is the real strategic value of this deal.
Former Department of Energy Loan Programs Office head Jigar Shah's assessment is blunt: applying NextEra's energy storage expertise to Virginia's data center load 'could be transformative' — because data centers don't just need electricity; they need stable electricity, predictable electricity, and ideally electricity that can be stored during off-peak hours.
NextEra is betting that the demand for AI computing power won't stop.
Judging from current investment trends, this bet isn't aggressive. Through 'Large Load Tariff' mechanisms, major electricity consumers (i.e., data centers) will directly participate in funding infrastructure construction. This means NextEra can partially shift the capital pressure for future transmission line and generation facility expansions onto the tech companies — rather than bearing it alone as a utility.
Of course, regulatory challenges also lie ahead.
Acquiring Dominion means NextEra will become a multi-state electricity super-giant, likely facing intense scrutiny from state public utility commissions across its territory. Consumer advocacy groups like Clean Virginia have already publicly warned, calling for the 'strictest scrutiny' of the deal, fearing the concentration of control over Virginia's power grid.
04 Who Pays the Power Bill?
When electricity resources are being voraciously consumed by AI and power bills soar, the ultimate question is: who ends up paying for the increased cost of electricity? This might be the most important question behind this monumental acquisition.
Electricity infrastructure construction requires money, and that money ultimately finds its way into electricity prices through various means. Some U.S. utilities have already started using 'Construction Work in Progress' financing mechanisms, allowing them to charge consumers for projects before they are completed. In other words, residential users are already paying for data center infrastructure construction before they see any benefit from the increased grid capacity.
PowerLines analysis provides a startling figure: residential consumers may bear about $700 billion of the costs for AI-driven electricity infrastructure investment, gradually transferred through higher power bills.
$700 billion. This is on the same order of magnitude as the capital expenditures of tech companies, but the flow is completely different. The tech companies' $400 billion in capital expenditure brings shareholder returns, improved model capabilities, and corporate competitive advantage. The portion of costs borne by consumers, however, only buys them a steeper curve on their electricity bill.
There is a structural unfairness written into the logic of this merger, and into the entire wave of AI infrastructure investment.
Data centers are private assets; the economic benefits of AI are concentrated in the hands of tech companies and their shareholders. But the grid that supports all this operation is public infrastructure; its construction and maintenance costs are shared by all users. This isn't a new problem, but AI has magnified it to an unprecedented scale.
The $67 billion acquisition lays bare, for the first time so clearly to everyone, the consolidation logic of the energy industry: The prosperity of AI doesn't only happen inside data centers. It spreads outward along power cables, into the grid, onto the balance sheets of utility companies, and ultimately into the electricity bill of every ordinary household.
This merger is not the end. Given the current rate of AI computing power expansion, this is likely just a beginning — the reshuffling of the power grid landscape has only just begun.
