UK National Grid Invests in US Power Companies Amidst AI Electricity Demand Surge

TL;DR · BNEF predicts that electricity demand from US data centers could reach 106GW by 2035. · New York State has paused environmental permits for some ultra-large data centers, leading to market trading of electricity access constraints. · Related sectors: utilities, nuclear energy, grid equipment, copper, data center REITs.

In recent months, discussions around AI infrastructure in the US have expanded from chip supply to electricity access. BNEF has raised its electricity demand forecast for data centers, while National Grid Ventures invests in large load power projects in the US, and New York State has paused some new ultra-large data center environmental permits.

These three clues point to the same question: after AI companies acquire GPUs, can they still secure land, cooling, water, grid access, and long-term power supply as planned?

For investors, data centers are no longer just capital expenditure projects for tech companies. The electricity consumption of large campuses could approach that of a small city, and as AI training and inference continue to expand, utilities, gas, nuclear energy, grid equipment, copper, and data center REITs will all be included in the AI pricing chain.

Electricity Demand Forecast Raised, Grid Nodes Under Pressure

In its report at the end of 2025, BNEF predicts that electricity demand from US data centers could reach 106GW by 2035, a 36% increase from its forecast seven months ago. This is not the already realized electricity consumption but a long-term demand benchmark for the market.

106GW can be understood as the capacity of a group of large power plants operating at full load over the long term. The more troubling aspect is that this load will not be evenly distributed across the US but concentrated in a few states and grid regions with data centers.

BNEF also predicts that the capacity of data centers in the PJM grid region could reach 31GW by 2030. PJM covers several states in the eastern US and is one of the power markets where data centers and industrial loads are concentrated. When load grows concentrated, the issue shifts from whether national power generation is sufficient to whether local nodes can handle the load.

This explains why electricity demand forecasts can impact asset pricing. Buying GPUs is just the first step for AI companies; putting data centers into operation requires power sources, transformers, transmission lines, backup power, and long-term power purchase agreements. Compared to servers, grid access and permits are harder to replicate quickly.

Power Capital Begins to Bind AI Loads

On July 1, National Grid Ventures announced it had acquired a 35% stake in Joulent for $1.75 billion, participating in large load power infrastructure projects in the US. This transaction signals that traditional power capital is beginning to view AI data centers as long-term load assets.

Joulent's first project, Project Kilby, is located in West Texas and is a 2.67GW supporting power facility, with Chevron holding a 50% stake. The project plans to serve data centers operated by Microsoft through a 20-year power purchase agreement, aiming to start supplying power in 2028.

The focus is not on which power source route prevails, but rather that data centers are shifting from "waiting for power" to "locking in power in advance." Long-term power purchase agreements and self-supplied power are becoming prerequisites for computational power expansion.

The logic of self-supplied power is straightforward. If data centers continue to draw power from the public grid, the expansion costs may be spread across all users. If project developers build their own power sources or sign long-term contracts, they can improve supply certainty and more easily prove to regulators that they will not crowd out residential electricity.

Risks remain. Gas projects face fuel price and emission pressures, nuclear projects face regulatory and construction cycle challenges, and grid expansion is constrained by transformers, transmission lines, and local permits. The binding of capital to power sources in advance indicates that demand is being taken seriously and that bottlenecks have become sufficiently specific.

New York Pushes Cost Allocation to the Forefront

On July 14, New York Governor Kathy Hochul signed an executive order pausing state environmental permits for new ultra-large data centers for up to a year. More precisely, the pause applies to related permit applications that have not yet been fully recognized, not all data center constructions.

The state government cited reasons including protecting consumers, the environment, the grid, and communities, while also considering the impacts on water resources and electricity costs. This has created a clear constraint in the AI electricity narrative: data centers can expand, but they cannot leave the burden of grid upgrades, water resource pressures, and rising residential electricity costs to local communities.

New York State also stated that the Energize NY program will require data centers to pay higher energy costs or provide their own power, and will consider grid acceleration funds and dedicated clean power requirements. This is not a simple opposition to AI but a demand for high-energy-consuming projects to internalize their true costs.

This will change the competitive dimensions for data center operators and REITs. In the past, the market valued land, leases, customer quality, and financing ability. Now, it will also consider whether projects can secure electricity, obtain permits, and prove they will not raise residential electricity prices.

New York may not represent the entire US. However, if similar pressures arise in high-load areas like Virginia, Georgia, and Texas, the timeline for AI infrastructure expansion will shift from "who buys chips first" to "who secures local permits and verifiable power sources first."

Nuclear Energy Provides Long-Term Options

The Trump administration's push for advanced reactors, AI data centers, and federal site deployments in 2025 has brought nuclear energy back into the AI electricity narrative. For data centers, the appeal of nuclear energy lies in its stability, low carbon emissions, and long-term power supply.

Companies like Oklo, X-Energy, Aalo Atomics, Valar Atomics, and Helion have garnered market attention due to this vision. Investors hope to find power solutions that are more certain than traditional grids and more stable than intermittent renewable energy.

However, policy support does not equate to commercial delivery. Traditional nuclear power construction cycles are long, and advanced reactors face challenges related to regulation, supply chains, fuel, financing, and public acceptance. Even if approvals speed up, they cannot deliver tens of GW of power to data centers in the short term.

Thus, nuclear energy resembles a long-term pricing option. It explains why nuclear companies, uranium, engineering services, and grid equipment have been included in the AI trading chain, but it does not prove that the AI electricity bottleneck has a definitive solution. Equating policy catalysis directly with order fulfillment is a major risk in this narrative.

Cost Realization Determines Market Dynamics

Whether AI electricity trading can shift from thematic trading to performance trading depends on two variables: whether data centers can secure electricity as planned, and whether new costs can be clearly allocated.

If self-supplied power and long-term power purchase agreements can be scaled effectively, utilities, grid equipment, gas, and some nuclear assets will gain clearer order anchors. AI companies can also exchange higher electricity costs for certainty in computational power expansion.

If local moratoriums spread, or residential electricity cost pressures become a political issue, the pace of data center construction may be extended. The pressure may not necessarily be on AI demand itself, but on the site selection, grid connection, and profit margin assumptions of high-energy-consuming projects.

The core of this main line is not that "AI will definitely lack electricity," nor is it that "nuclear energy will solve the problem immediately." A more tradable judgment is that AI expansion is turning the electricity system into a new pricing anchor. Those who can prove they bring electricity are more likely to convert computational demand into revenue and valuation.

Disclaimer: This content is provided for general branding and informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online events, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets or to use any services. Crypto assets are highly volatile and may result in loss. WEEX services and online events may not be available in all regions and are subject to applicable laws, regulations, and eligibility requirements. You are responsible for ensuring that your use of WEEX services complies with local laws and for carefully assessing the risks before participating in any crypto-related activities.