Paducah AI campus turns former uranium site into $100bn infrastructure bet

Brookfield Asset Management Ltd. (NYSE: BAM, TSX: BAM) and NextEra Energy, Inc. (NYSE: NEE) are leading a coalition planning a privately funded $100 billion artificial intelligence data centre and energy complex at the United States Department of Energy’s former Paducah uranium-enrichment site in western Kentucky. The proposed Paducah American Energy Hub would support more than 1.2 gigawatts of computing capacity and as much as 1.8 gigawatts of utility capacity when fully developed in 2032. NextEra Energy plans to develop up to 2 gigawatts of natural gas-fired generation and 2.6 gigawatts of battery energy storage, while Brookfield would lease the federal land and develop and operate the data centre campus. The project is strategically significant because it attempts to solve artificial intelligence infrastructure’s power, land, grid and financing constraints within one coordinated development. The announcement remains subject to definitive agreements, utility arrangements and regulatory approval, meaning the headline investment is a development ambition rather than committed construction expenditure.

Why does the Paducah AI data centre project combine computing capacity with 4.6 gigawatts of dedicated energy resources?

Artificial intelligence data centres are becoming power projects with server buildings attached. The largest proposed campuses can require as much electricity as industrial cities, making conventional utility interconnection timelines increasingly incompatible with hyperscalers seeking to deploy computing capacity within two or three years.

The Paducah project attempts to address that problem by developing the energy supply alongside the data centre. NextEra Energy would add natural gas generation in stages as Brookfield brings computing capacity online, while battery systems would support grid balancing, peak demand management and reliability.

The proposed 4.6 gigawatts of generation and storage exceeds the campus’s stated 1.8 gigawatts of utility capacity. The difference reflects the fact that battery storage cannot be treated as continuous generation, while natural gas plants may not operate at maximum output every hour. The additional capacity also creates the possibility of supplying electricity to the wider regional grid when the data centre is not using the full system.

This architecture is intended to reduce dependence on existing generation and prevent the project from consuming electricity already allocated to households and businesses. The companies are presenting the campus as capable of paying for its own power infrastructure rather than requiring local ratepayers to finance transmission and generation upgrades.

That promise addresses one of the most politically sensitive issues surrounding data centre development. Communities increasingly question why residential customers should absorb higher grid costs when technology companies consume extraordinary quantities of electricity. The Paducah structure seeks to separate the project’s infrastructure burden from the existing customer base.

The economic reality will depend on the final power service agreement. Costs can still move indirectly through transmission upgrades, reliability arrangements and shared utility assets. Regulators will need to determine whether the contracts genuinely isolate customers from financial risk or merely allocate that risk through more complicated mechanisms.

Natural gas gives NextEra Energy a dispatchable source capable of operating when artificial intelligence workloads require continuous electricity and renewable output is unavailable. Battery storage can respond quickly to changes in demand but cannot replace fuel-based generation during a prolonged period without sufficient charging supply.

The project therefore reflects a pragmatic rather than purely renewable approach to artificial intelligence power demand. It combines natural gas, storage and grid access instead of depending on one technology. That may improve reliability, but it will also attract scrutiny over carbon emissions and the long-term economics of building new gas assets for a rapidly changing computing market.

Why was the former Paducah uranium-enrichment site selected for a hyperscale computing campus?

The Paducah Gaseous Diffusion Plant supported the United States atomic energy programme for decades before enrichment operations ended in 2013. Its industrial legacy left the site with large tracts of federal land, transmission infrastructure, roads, water access and fibre connectivity.

Those characteristics are increasingly scarce near established data centre markets. Northern Virginia, Northern California and other major hubs face constraints involving land, transmission capacity, permitting and community opposition. A former industrial site offers fewer competing land uses and infrastructure originally designed for energy-intensive operations.

The United States Department of Energy selected Paducah as one of several federal properties suitable for artificial intelligence data centres and supporting generation. The government requested privately funded proposals in November 2025, requiring applicants to build, operate and eventually decommission the infrastructure while securing their own interconnection and regulatory approvals.

Brookfield was selected to lease land and develop the computing campus, while NextEra Energy was selected to build and own the supporting energy assets. Big Rivers Electric Corporation would provide wholesale power service, Jackson Purchase Energy Cooperative would deliver retail service and Paducah Power System would participate as a community partner.

Reusing federal land can shorten part of the development process because land ownership is already consolidated and major infrastructure exists. It does not eliminate environmental, permitting or construction risks. The site remains part of a long-running environmental cleanup associated with its former enrichment operations.

That history creates an unusual contrast. Land once central to nuclear weapons and fuel production could now support the computing infrastructure viewed as strategically important to competition with China. The underlying policy logic remains similar, even though centrifuges and gaseous diffusion equipment are giving way to graphics processors and cooling systems.

The project could also generate a productive use for land that might otherwise remain economically constrained for decades. Approximately 8,000 construction jobs and 600 permanent operational positions are expected if the campus reaches full scale.

The permanent employment figure is relatively modest compared with the headline investment because data centres are highly capital intensive. The wider economic case will depend on construction activity, local procurement, tax receipts, power infrastructure and whether additional technology or manufacturing businesses cluster around the site.

Can Brookfield commit $100 billion before the project has disclosed its hyperscale customers?

The $100 billion figure represents the planned private investment across the broader computing and infrastructure buildout rather than capital already transferred into the project. Brookfield has described Paducah as the starting point for a wider plan to deploy $100 billion into artificial intelligence infrastructure.

A substantial portion of the eventual expenditure would likely involve computing equipment purchased or financed for future tenants. Graphics processors, networking systems, storage, servers and cooling technology can account for a large share of an artificial intelligence campus’s total cost.

Brookfield is expected to provide capital, development expertise and access to institutional investors. The asset manager can structure separate vehicles involving infrastructure funds, credit investors, strategic partners and future data centre tenants rather than carrying the full investment on its own corporate balance sheet.

That model is increasingly common as hyperscale computing requirements exceed what even the largest technology companies want to finance directly. Asset managers can own the land, buildings and supporting infrastructure while technology tenants sign long leases or capacity agreements.

The arrangement allows hyperscalers to access computing capacity without immediately owning every physical asset. It also creates long-duration contracted cash flows that can appeal to pension funds, sovereign investors and insurance capital.

However, the Paducah announcement did not identify a committed technology tenant. That absence is the most important commercial gap in the project. A 1.2-gigawatt computing campus requires one or more customers with extraordinary capital resources and confidence in future artificial intelligence demand.

Brookfield may be negotiating with cloud providers, model developers or enterprises seeking dedicated artificial intelligence capacity. Until binding leases or capacity contracts are signed, the development remains exposed to utilisation risk.

The difference between a viable campus and an expensive collection of empty buildings is contracted demand. Brookfield can stage construction to limit speculative exposure, but the overall $100 billion figure assumes that major technology customers eventually commit.

This makes tenant disclosure a more valuable milestone than another conceptual partnership. Investors should watch for lease duration, credit support, commencement dates and whether computing equipment is funded by Brookfield, the tenant or a separate financing vehicle.

How does the Paducah project fit Brookfield Asset Management’s broader AI infrastructure strategy?

Brookfield Asset Management manages more than $1 trillion across infrastructure, renewable power, real estate, private equity and credit. Artificial intelligence infrastructure sits at the intersection of several of those strategies, allowing Brookfield Asset Management to participate through land, power generation, data centres, transmission, financing and operating platforms.

The Paducah campus fits a broader effort to position Brookfield Asset Management as a provider of physical infrastructure for artificial intelligence. The company has pursued data centre, power and computing opportunities across North America, Europe and Asia, including rapidly expanding markets such as India.

This breadth gives Brookfield Asset Management an advantage over a specialised data centre developer. The company can combine development capability with institutional capital and relationships across power, infrastructure and technology.

It can also structure transactions differently depending on the customer. One tenant may prefer a conventional lease, another may purchase capacity and a third may form a joint venture. Brookfield Asset Management can earn management fees and performance-related returns while limiting the amount of corporate capital directly exposed to each project.

First-quarter 2026 fee-related earnings reached $772 million, increasing 11% from a year earlier. Brookfield Asset Management raised $21 billion during the quarter and had attracted approximately $67 billion of capital during the early part of 2026, providing substantial fundraising capacity for large infrastructure strategies.

The asset-light structure means the listed company’s direct financial exposure may be smaller than the $100 billion headline suggests. Brookfield-managed funds and partners would likely provide much of the capital, while Brookfield Asset Management earns fees for raising, deploying and managing it.

That model protects the corporate balance sheet but does not eliminate reputational or fundraising risk. A poorly performing artificial intelligence infrastructure programme could weaken investor confidence in future Brookfield funds and reduce performance income.

The project also lengthens the duration between capital commitment and fee-generating deployment. Data centre campuses and power plants can take years to permit and construct. Brookfield Asset Management must raise capital without allowing undeployed commitments to sit idle for too long.

The strategic opportunity is significant because artificial intelligence could become one of the largest infrastructure investment cycles in decades. Brookfield Asset Management is attempting to own the toll roads, power systems and buildings rather than predict which model developer ultimately wins.

Why is NextEra Energy adding natural gas and battery storage instead of relying on one power technology?

NextEra Energy’s proposed contribution includes up to 2 gigawatts of natural gas generation and 2.6 gigawatts of battery storage. This mix reflects the different operating roles required to support a large artificial intelligence campus.

Natural gas plants can provide sustained electricity and respond to periods when renewable resources are unavailable. Battery systems can deliver power quickly, absorb excess supply and reduce sharp changes in grid demand as computing workloads fluctuate.

Artificial intelligence data centres require extremely high reliability. Interruptions can halt expensive training runs, damage equipment or prevent customers from meeting service commitments. Developers therefore need multiple layers of power supply, storage and backup.

A hybrid system also allows NextEra Energy to stage investment. Generation resources can be added as the campus signs tenants and increases computing capacity. That reduces the risk of building all 4.6 gigawatts before demand has materialised.

The project fits NextEra Energy’s expanding exposure to data centre power demand. The company operates Florida Power & Light Company and NextEra Energy Resources, giving it experience across regulated utility investment, renewable generation, natural gas, nuclear power and battery storage.

Second-quarter 2026 results showed the financial importance of rising electricity consumption. Florida Power & Light Company’s quarterly net income increased 10.2% to approximately $1.41 billion, while NextEra Energy Resources’ net income rose 66.2% to about $1.63 billion.

NextEra Energy Resources added approximately 3.6 gigawatts of storage projects during the quarter, taking its development backlog to around 35.1 gigawatts. Paducah would therefore become a large but strategically consistent addition to an existing storage and generation pipeline.

The project could generate revenue through long-term contracts rather than direct exposure to volatile electricity prices. Contract structure will determine whether NextEra Energy receives sufficiently predictable returns to justify the construction and financing risk.

The principal concern is customer certainty. The value of dedicated generation depends on the data centre operating as planned. NextEra Energy must avoid committing billions of dollars before creditworthy tenants provide adequate contractual support.

Can the project prevent local electricity customers from absorbing AI infrastructure costs?

The coalition has made ratepayer protection central to the announcement. The data centre is intended to pay for its dedicated generation, storage and supporting infrastructure, while excess power could strengthen the regional grid.

This model responds directly to public resistance emerging in several United States data centre markets. Residents have challenged new projects over electricity rates, water consumption, land use and the limited number of permanent jobs created relative to infrastructure demands.

The Paducah power service agreement will require oversight from the Kentucky Public Service Commission. Regulators will examine cost allocation, demand guarantees, interconnection expenses and the financial consequences if construction is delayed or tenants consume less electricity than expected.

A strong contract would require the project to pay even when it uses less power than forecast. Without that protection, utilities and cooperative members could be left supporting infrastructure built for a customer that never reaches full demand.

The agreement must also address decommissioning and stranded assets. A natural gas plant can operate for decades, while artificial intelligence hardware cycles are measured in years. The economic life of the power assets may exceed the commercial life of the initial computing technology.

Excess power could benefit the region when the data centre does not consume the entire output. However, those benefits depend on transmission capacity and regional market conditions. Additional electricity does not automatically reduce customer bills when network constraints prevent it from reaching demand centres.

Battery storage can improve reliability by responding during peak periods or system disruptions. Yet batteries require charging electricity and eventually need replacement. Their full lifecycle cost must be reflected in the customer contracts.

The project’s ratepayer promise is therefore credible in concept but unproven in execution. The Kentucky Public Service Commission’s eventual decision will reveal whether financial risk is genuinely assigned to Brookfield, NextEra Energy and future tenants.

Other states will watch closely. A successful structure could become a template for pairing artificial intelligence data centres with dedicated generation. A poorly allocated agreement could become evidence that “bring your own power” is more slogan than protection.

What regulatory, environmental and construction risks could delay the Paducah American Energy Hub?

The project remains subject to negotiation and execution of definitive documentation. This means land leases, power service arrangements, construction contracts, tenant commitments and financing structures have not all reached final form.

The Kentucky Public Service Commission must review the utility arrangements. Natural gas generation will require permits, fuel supply, pipeline access and environmental approvals. Battery projects need land, interconnection studies, fire-safety planning and equipment procurement.

The former uranium-enrichment site adds environmental complexity. Although the project would use land selected for redevelopment, construction must remain coordinated with the Department of Energy’s continuing cleanup and decommissioning programme.

Developers must demonstrate that contamination does not create unacceptable worker, construction or operational risk. Environmental monitoring and site controls could affect which parcels are available and how quickly ground preparation proceeds.

Supply-chain constraints present another risk. A 1.2-gigawatt computing campus could require enormous quantities of transformers, switchgear, turbines, batteries, fibre equipment and cooling systems. Many of these components already have extended delivery timelines.

The project also needs a high-capacity natural gas connection. Securing pipeline capacity can be politically and technically difficult, particularly when regional infrastructure was not designed for a new two-gigawatt generator.

Technology risk will evolve during construction. Artificial intelligence chips are becoming more powerful and energy dense, changing rack layouts, cooling methods and electrical architectures. Buildings designed too early around current hardware could require expensive modifications before completion.

The planned 2032 completion date provides time for staged development but also exposes the campus to several hardware generations and potential changes in artificial intelligence economics. Demand may increase dramatically, or efficiency improvements may reduce the amount of physical infrastructure required for each unit of computing.

Execution discipline will therefore matter more than the announced maximum capacity. Brookfield and NextEra Energy should expand only when tenant commitments, power requirements and technology specifications justify the next phase.

What does the latest stock performance indicate about investor sentiment toward Brookfield and NextEra Energy?

Brookfield Asset Management shares were trading near $47.84 at approximately 1:51 p.m. Eastern Time on July 30, up around 1% during the session. The stock had gained approximately 4.7% from its July 23 close and 6.7% from its June 30 close.

Brookfield Asset Management remained within a 52-week range of approximately $42.20 to $64.10. The July 30 intraday price was about 25% below the high, indicating that investor sentiment has improved recently but remains well below the enthusiasm reflected in the previous peak.

The positive movement cannot be attributed solely to the Paducah announcement. Brookfield Asset Management’s valuation is influenced by fundraising, fee-related earnings, asset realisations, interest rates and the outlook for alternative investments across several strategies.

The project supports the company’s long-term infrastructure narrative, but investors still need evidence of tenant commitments and fee-generating capital deployment. A $100 billion plan can create management fees only when capital is raised and invested.

NextEra Energy shares were trading near $87.49 at approximately 1:51 p.m. Eastern Time on July 30, down about 1.1% during the session. The stock was approximately 2.6% below its July 23 level and broadly flat, down around 0.3%, compared with the end of June.

NextEra Energy remained within a 52-week range of approximately $69.24 to $98.75. The latest price was roughly 11.4% below the high and 26.4% above the low.

The muted market response suggests investors are treating Paducah as an early-stage development opportunity rather than a secured earnings catalyst. The lack of named data centre customers, final agreements and detailed capital requirements limits the amount of near-term value that can reasonably be assigned.

Investor sentiment toward NextEra Energy remains supported by second-quarter earnings, regulated utility growth and the company’s large development backlog. However, markets are becoming more selective about the artificial intelligence power trade as project announcements multiply faster than confirmed computing tenants.

The next valuation catalysts will be definitive contracts, regulatory approval, tenant disclosure and clarity on NextEra Energy’s expected investment and returns. Until then, the project strengthens strategic positioning without materially changing near-term earnings expectations.

What milestones will show whether the $100 billion Paducah AI campus is becoming a real project?

The first milestone will be signed definitive agreements between the United States Department of Energy, Brookfield, NextEra Energy and the participating utilities. The current announcement establishes the intended structure but does not complete the commercial arrangements.

The second milestone will be the identification of one or more creditworthy data centre tenants. Their lease commitments will determine how quickly Brookfield can finance and construct the campus.

The third milestone will be Kentucky Public Service Commission approval of the power service structure. The decision will show whether regulators accept the claim that local customers are protected from project costs.

The fourth milestone will be natural gas supply and interconnection agreements. A two-gigawatt generation plan requires fuel transportation and transmission capacity on a scale that cannot be assumed.

The fifth milestone will be an initial construction phase with defined compute, power and storage capacity. A smaller first phase would demonstrate that the companies are matching capital deployment with contracted demand.

The sixth milestone will be disclosure of financing arrangements. Investors need to know how much capital will come from Brookfield-managed funds, NextEra Energy, tenants, debt providers and other strategic partners.

The seventh milestone will be evidence that cleanup and data centre development can proceed simultaneously without creating environmental or scheduling conflicts.

The eighth milestone will be the first operating capacity expected before full completion in 2032. Actual electricity consumption, server utilisation and tenant expansion will provide a better measure of demand than the maximum campus design.

Paducah has the land, infrastructure, capital sponsors and national-security narrative required for a major artificial intelligence hub. What it does not yet have publicly is the customer commitment that converts those advantages into an investable project.

What are the key takeaways from the $100 billion Paducah AI infrastructure plan?

• Brookfield and NextEra Energy are leading a proposed $100 billion privately funded artificial intelligence data centre and energy campus in western Kentucky.
• The Paducah American Energy Hub could support more than 1.2 gigawatts of compute capacity and up to 1.8 gigawatts of utility capacity by 2032.
• NextEra Energy plans up to 2 gigawatts of natural gas generation and 2.6 gigawatts of battery storage dedicated to the development.
• Brookfield would lease federal land and develop and operate the data centre campus through institutional infrastructure capital.
• The former uranium-enrichment site offers existing transmission, roads, water infrastructure, fibre connectivity and large parcels of industrial land.
• No hyperscale technology tenant has been publicly identified, leaving customer commitment as the project’s largest commercial uncertainty.
• The power model is intended to protect existing ratepayers, but the Kentucky Public Service Commission must test whether costs and stranded-asset risks are genuinely isolated.
• Brookfield Asset Management shares were up approximately 4.7% over five trading sessions and 6.7% over one month at the July 30 intraday price.
• NextEra Energy shares were down approximately 2.6% over five trading sessions and broadly flat over one month as investors awaited firmer project economics.
• Definitive agreements, named tenants, regulatory approval and a financed first construction phase will determine whether the $100 billion vision becomes operating infrastructure.