National Grid Pays $1.75 Billion for a 35% Stake in Joulent — and a Seat at the AI Power Table

The question of who builds the power stack underneath AI has a new, significant answer. Britain's National Grid has committed $1.75 billion for a 35% stake in Joulent, a U.S. energy platform purpose-built for data center power infrastructure. The deal isn't a passive portfolio move — it's a structural bet that the electricity bottleneck constraining AI expansion is the most consequential infrastructure problem of the decade.

What Joulent Actually Does

Joulent's entire value proposition is pointed at one problem: connecting high-load AI and cloud facilities to reliable, scalable power. As AI clusters grow denser and more power-hungry, conventional grid interconnection timelines and capacity constraints have become a hard ceiling on deployment speed. Joulent's platform targets that ceiling — developing the power infrastructure that sits between the transmission grid and the data center fence.

The company isn't a traditional utility and isn't trying to be. It operates in the space where grid operators, hyperscalers, and capital markets intersect — financing and planning AI-oriented energy projects at the pace the market demands, not at the pace of legacy utility procurement cycles.

Why National Grid Is Buying In Now

For a British transmission operator to write a $1.75 billion check into a U.S. AI infrastructure play requires a particular read of where durable returns are forming. National Grid has framed this explicitly as part of its broader strategy to support high-load AI and cloud facilities in the United States — not as a general grid expansion play, but as a deliberate position in a specific, high-growth segment.

The 35% stake gives National Grid more than financial exposure. It grants significant influence over how Joulent plans and finances future AI-oriented energy projects — who gets capacity, on what terms, and in which markets. That kind of structural leverage over pipeline decisions is worth considerably more than a minority equity return in a rising sector.

The timing matters too. This investment registers as one of the largest traditional-utility capital commitments to AI-specific power infrastructure to date. Utilities entering this space now are shaping the terms of the market before demand outpaces the available financing and development capacity.

The Signal for Builders and Operators

For founders and operators building in the AI infrastructure stack, the National Grid–Joulent deal is a data point worth sitting with. The capital needed to build AI data center power capacity at scale is moving from venture balance sheets toward large institutional players with long time horizons and tolerance for complex energy project finance. That changes the competitive dynamics of who can move fast and who gets access to power.

It also signals that the power bottleneck is being treated as a first-order problem by some of the largest capital allocators in infrastructure — not a second-order concern to be solved incrementally. When a national transmission operator structures a deal specifically around AI electricity demand rather than conventional load growth, it reflects a view that the demand curve is real, durable, and large enough to underwrite multibillion-dollar commitments.

The Bigger Shift

The deeper story here isn't National Grid diversifying into U.S. markets. It's that the electricity infrastructure layer beneath AI is attracting the same quality of institutional conviction — and capital scale — that the compute layer attracted two years ago. Joulent's model, and National Grid's willingness to pay $1.75 billion for a meaningful stake in it, suggests the industry has internalized that AI's next constraint isn't chips or models. It's watts. The companies that own the infrastructure connecting those watts to the machines will occupy a structurally advantaged position for a long time. National Grid just bought its way into that position.