National Grid shares fall as $1.75bn Joulent bet targets AI data centre power boom

National Grid plc (LSE: NG.) has agreed to invest $1.75 billion for a 35% interest in Joulent LLC, a United States energy platform developing power infrastructure for data centres and other large electricity users. Joulent’s first major development is Project Kilby, a 2.67-gigawatt West Texas power facility intended to supply a Microsoft-operated data centre campus under a 20-year power purchase agreement. The investment gives National Grid exposure to higher-return contracted infrastructure outside its traditional regulated networks, but it also adds a substantial capital commitment alongside the company’s existing £70 billion investment programme. National Grid shares fell 2.4% to around 1,218 pence on July 1, indicating that investors are balancing the artificial intelligence electricity opportunity against higher execution and funding risks.

Why is National Grid investing $1.75 billion in Joulent rather than only expanding regulated networks?

National Grid’s core business earns regulated returns by investing in electricity and gas networks across the United Kingdom and northeastern United States. This model offers relatively predictable revenue, but returns are constrained by regulatory settlements and the pace at which permitted investment can be added to the company’s asset base. Joulent provides access to a different model built around privately contracted generation, high-voltage infrastructure and long-term agreements with large corporate customers.

The strategic attraction is the rapid growth in electricity demand from data centres supporting artificial intelligence, cloud computing and other compute-intensive activities. Data centre power consumption associated with generative artificial intelligence increased 17% in 2025, considerably faster than the 3% growth recorded across total global electricity demand. Traditional grid connection processes can take several years, creating a commercial opening for developers able to deliver dedicated generation and connection infrastructure more quickly.

National Grid is therefore purchasing exposure to both electricity demand growth and the scarcity of available power. Joulent is designed to offer integrated solutions that may include co-located gas generation, battery storage, renewable power and private connections linking large customers to the wider grid. The platform can potentially earn higher returns than regulated networks because it accepts greater construction, customer and market risk.

The investment also diversifies National Grid’s geographical presence within the United States. Most of National Grid’s regulated American operations are concentrated in New York and New England, while Project Kilby is located in West Texas. Texas offers strong electricity demand growth, substantial natural gas availability and a large pipeline of data centre development, but it also operates within a competitive power market that differs significantly from National Grid’s northeastern utility territories.

This is not a defensive infrastructure investment. National Grid is intentionally accepting more risk in exchange for access to a faster-growing market and potentially stronger returns. The transaction will be judged by whether those additional returns adequately compensate shareholders for construction complexity, commodity exposure and the long wait before Joulent becomes cash generative.

How could Project Kilby solve the speed-to-power problem facing United States data centres?

Project Kilby is being developed through a 50-50 partnership between Joulent and Chevron Corporation. The 2.67-gigawatt facility will provide dedicated power to a Microsoft-operated data centre campus in West Texas under a 20-year agreement, creating a large contracted customer base before construction is completed. Critical equipment, including GE Vernova turbines, has been secured, while engineering and construction capacity has also been reserved. Initial power delivery is targeted for 2028.

The scale is considerable. A 2.67-gigawatt facility represents power infrastructure comparable with several conventional utility-scale generating units combined. Building it next to or close to the customer reduces dependence on waiting for every component of the wider transmission system to be expanded before the data centre can begin operating.

This structure addresses one of the biggest constraints facing the artificial intelligence industry. Technology companies can procure chips, construct buildings and deploy computing systems faster than utilities can sometimes expand transmission networks. Power availability has consequently become a factor determining where data centres are built and how quickly new computing capacity can enter service.

Joulent’s model effectively places generation development, electrical infrastructure and the customer contract within the same commercial platform. That can simplify decision-making and reduce coordination delays involving utilities, generators, transmission owners and data centre developers.

National Grid contributes capabilities that are difficult for a newly created energy platform to build quickly. These include high-voltage network design, system integration, power balancing, infrastructure construction and experience working with regulators and local stakeholders. Chevron brings energy project execution and access to natural gas expertise, while Microsoft provides a creditworthy anchor customer.

However, securing turbines and reserving engineering capacity do not eliminate development risk. Project Kilby still requires a final investment decision, permitting, financing, construction and commissioning. Labour shortages, equipment delays, inflation and changes in data centre requirements could affect cost or timing before the targeted 2028 delivery date.

Can Microsoft’s 20-year power agreement make a higher-risk project behave like a utility asset?

The 20-year power purchase agreement is central to the investment case because it provides long-term demand visibility. Rather than constructing generation and hoping that wholesale electricity prices remain attractive, Project Kilby is being developed around a defined customer with substantial and relatively predictable power requirements.

Long-duration contracts can make commercial power projects resemble regulated infrastructure by reducing exposure to short-term electricity price movements. They can also support project financing because lenders and investors have greater confidence that the facility will generate revenue after entering operation.

Microsoft’s involvement strengthens the credit profile, but customer concentration remains an important consideration. The foundational project depends heavily on one data centre campus and the continued expansion of artificial intelligence infrastructure. Contractual protections may reduce demand risk, although the economic value of those protections will depend on pricing terms, performance requirements and termination provisions that have not been publicly disclosed.

Technology risk must also be considered. Artificial intelligence demand is currently increasing rapidly, but computing hardware is becoming more energy efficient. Future data centre architecture could change power requirements, cooling needs or the preferred locations of large campuses. A 20-year agreement reduces National Grid’s direct exposure to those changes, but future Joulent projects may not secure identical contractual terms.

The relationship could nevertheless create further opportunities. National Grid expects to connect more than 10 gigawatts of data centre demand across the United Kingdom and United States over the next five years. Experience gained through Project Kilby could improve how National Grid designs grid connections, evaluates large-load requests and works with technology companies across its regulated networks.

The transaction therefore offers strategic knowledge as well as financial exposure. National Grid will gain a closer view of how hyperscale customers plan electricity demand, manage reliability and evaluate private power solutions. That intelligence could become valuable as data centre developers compete for increasingly scarce grid capacity.

What does gas-fired generation mean for National Grid’s energy transition positioning?

Project Kilby’s foundational generation will be gas-fired, creating an unavoidable tension within National Grid’s wider energy transition narrative. The company is investing heavily in electricity transmission and distribution infrastructure needed to connect renewable generation, but the immediate solution for artificial intelligence demand in Texas relies on conventional gas generation.

The commercial reasoning is straightforward. Data centres require continuous electricity and cannot depend entirely on variable wind or solar output without substantial storage or backup capacity. Gas turbines can provide firm power, respond to changing demand and be delivered more quickly than some alternative baseload technologies.

Joulent’s broader model includes battery storage and renewable integration, which could reduce emissions intensity over time. The platform also intends to connect facilities to the wider grid eventually, meaning co-located generation may function as a bridge while network infrastructure catches up with customer demand.

However, a 2.67-gigawatt gas project represents a long-lived physical asset. National Grid and its partners must consider future carbon costs, emissions regulation, fuel supply and the possibility that technology customers impose stricter clean-energy requirements during the contract period.

The structure may attract scrutiny if dedicated gas plants are perceived as allowing data centre developers to expand faster while transferring environmental consequences to surrounding communities. Joulent will need to demonstrate that projects do not weaken local electricity reliability, raise consumer costs or create infrastructure burdens that are ultimately socialised.

National Grid’s credibility will depend on presenting a realistic pathway for combining firm generation with storage, renewables and eventual grid connection. Investors are likely to tolerate gas as part of a transitional reliability solution more readily than as an open-ended strategy for powering artificial intelligence expansion.

Can National Grid fund Joulent alongside its £70 billion investment programme without stretching the balance sheet?

National Grid has stated that the $1.75 billion investment is incremental to its existing programme of at least £70 billion through the 2031 financial year. The company expects to fund the transaction from available balance-sheet capacity without changing its current five-year financial framework or compromising its investment-grade credit position.

The funding claim is credible, but it does not make the investment financially insignificant. National Grid ended March 2026 with net debt of £44.2 billion after record annual capital investment of £11.6 billion. Group capital expenditure is expected to approach £13 billion in FY2027, while net debt is forecast to increase by slightly more than £6 billion as infrastructure spending accelerates.

National Grid’s regulated asset growth provides support because approved investment eventually expands the revenue-producing asset base. The company expects assets to grow around 10% annually and underlying earnings per share to increase by 8% to 10% a year across the five-year framework. FY2027 underlying earnings per share are expected to rise between 13% and 15% from the FY2026 baseline of 78 pence.

Joulent has a different cash profile. The platform is not expected to become free cash flow positive until the early 2030s, meaning National Grid must fund its participation for several years before receiving meaningful cash returns. Any additional projects within Joulent’s multi-gigawatt pipeline could require further capital beyond the initial commitment.

The platform could eventually fund new projects internally or distribute cash to its investors. Before that point, National Grid remains exposed to development spending and potential cost increases. Investors will want clear disclosure on how quickly the $1.75 billion is deployed, whether future contributions are capped and what happens if Project Kilby exceeds its current budget.

The deal does not threaten National Grid’s financing position on its own. The concern is cumulative. A utility already undertaking one of the largest capital programmes in its history has less room for error when adding commercial projects that sit outside conventional regulated recovery mechanisms.

How could Joulent reshape competition for artificial intelligence electricity infrastructure?

The data centre electricity market is attracting utilities, oil and gas producers, renewable developers, private infrastructure funds and specialist power companies. Each group is attempting to solve the same problem: technology companies require power faster than conventional grid expansion can often provide it.

National Grid’s partnership with Chevron demonstrates how those capabilities are converging. A grid company contributes electrical engineering and transmission expertise, an energy producer contributes generation and fuel capabilities, and a technology customer provides long-term demand.

This model could become increasingly common. Oil and gas companies can use their balance sheets and gas resources to enter the power market, while utilities can move beyond regulated connections into private infrastructure. Infrastructure funds may then provide project capital once construction and contracting risks have been reduced.

National Grid possesses an advantage because it understands both sides of the connection process. It can help develop immediate private power solutions through Joulent while preparing the longer-term grid infrastructure needed to integrate those facilities into wider electricity systems.

The risk is that commercial partnerships begin competing with regulated networks for capital, equipment and skilled workers. Turbines, transformers, high-voltage cables and engineering capacity are already constrained in several markets. Projects backed by large technology customers may secure resources more quickly than conventional utility developments, potentially increasing costs elsewhere.

National Grid must also manage regulatory boundaries carefully. Its commercial activities should not benefit unfairly from information, relationships or infrastructure developed through regulated businesses. Transparent governance will be necessary to prevent concerns that utility customers are indirectly supporting higher-return private projects.

Why did National Grid shares fall despite the long-term artificial intelligence power opportunity?

National Grid shares finished July 1 near 1,218 pence, down approximately 2.4%, compared with a decline of about 0.4% for the FTSE 100. The larger fall indicates that the announcement created company-specific concerns rather than merely reflecting a weak market session.

The shares were approximately 1.9% below their June 24 closing level of 1,241 pence but remained around 2.9% above the June 1 close of 1,184 pence. National Grid was trading within a 52-week range of approximately 1,000.5 pence to 1,428.5 pence, placing the stock about 15% below its annual high and 22% above its low.

This performance suggests that sentiment remains broadly constructive but sensitive to capital allocation. Investors value National Grid partly for predictable regulated earnings and dividends. A $1.75 billion commitment to a private platform introduces a risk profile that some income-focused shareholders may not have expected.

The transaction also arrives when National Grid’s debt and investment requirements are already rising. Even though management expects no change to the financial framework, investors may apply a higher risk discount until the company discloses more information on project economics and capital commitments.

At around 1,218 pence, National Grid had a market capitalisation near £60.6 billion and a prospective dividend yield close to 4%. The valuation reflects a combination of defensive utility income and significant regulated growth, meaning investors do not necessarily need higher-risk commercial projects for the equity story to work.

Joulent must therefore deliver more than revenue growth. It must produce returns clearly above regulated alternatives without weakening dividend growth, credit quality or the funding of National Grid’s core networks.

What must happen before National Grid commits final capital and Project Kilby starts in 2028?

The final investment decision is expected before the end of 2026. National Grid and its partners must complete remaining commercial, regulatory, engineering and financing work before committing the full construction capital required for Project Kilby.

Investors should look for clearer disclosure on the project budget, National Grid’s maximum exposure and the return assumptions embedded within the investment. The company has said the opportunity can exceed regulated returns, but the financial premium must be substantial enough to justify the additional risks.

Permitting and construction milestones will provide the next evidence of execution. Securing GE Vernova turbines and reserving engineering capacity reduces supply-chain uncertainty, but the partners must still deliver a multi-gigawatt facility on an accelerated timetable.

National Grid must also explain how storage, renewable generation and future grid connectivity will be incorporated. A credible decarbonisation pathway would reduce regulatory and reputational risk while making the platform more attractive to technology customers with long-term clean-energy commitments.

Joulent’s wider development pipeline will be another important indicator. Project Kilby provides a significant anchor, but the $1.75 billion valuation case depends partly on the platform producing repeatable projects rather than remaining dependent on a single Microsoft campus.

National Grid has identified one of the largest emerging bottlenecks in the digital economy. Artificial intelligence needs electricity, and electricity infrastructure cannot be produced with a software update. The strategic opportunity is real, but National Grid must prove that moving faster than the conventional grid does not mean moving beyond its financial and operational discipline.

Key takeaways on what National Grid’s Joulent investment means for investors and the energy industry

• National Grid will invest $1.75 billion for a 35% interest in Joulent, gaining exposure to United States data centre electricity demand.
• Project Kilby will provide 2.67 gigawatts of dedicated power to a Microsoft-operated campus under a 20-year agreement.
• The project is being developed with Chevron and is targeting first electricity delivery in 2028.
• Long-term contracted revenue reduces power-price exposure but does not eliminate construction, customer concentration or technology risks.
• National Grid expects Joulent returns to exceed regulated-network returns because the platform carries a higher risk profile.
• Gas-fired generation provides reliable power quickly but creates carbon and energy-transition questions for National Grid.
• The investment is additional to National Grid’s £70 billion capital programme and comes as net debt is already expected to rise.
• Joulent is not expected to become free cash flow positive until the early 2030s, creating a lengthy funding period before returns emerge.
• National Grid shares fell 2.4%, suggesting that investors require more evidence on returns, capital limits and project execution.
• The final investment decision expected by the end of 2026 will be the next major test of the commercial and financial case.