The Nuclear Comeback

Nuclear Is Back in the Data Center Conversation

For much of the past decade, the data center industry's clean-energy strategy was dominated by renewable power.

Corporate buyers signed large wind and solar agreements. Power Purchase Agreements expanded rapidly. Sustainability targets pushed operators toward increasingly ambitious renewable procurement programs.

That strategy remains important.

But the energy conversation has expanded.

Microsoft is supporting the restart of a retired nuclear reactor in Pennsylvania. Google has committed to advanced reactors developed by Kairos Power. Amazon is investing in existing nuclear generation as well as Small Modular Reactors. Meta has signed agreements supporting existing nuclear plants while committing capital to future advanced nuclear projects.

These are not isolated announcements.

Taken together, they point to a larger market development:

Nuclear power has returned to the center of data center energy strategy.

The reason is not simply sustainability.

Data centers increasingly need energy that combines scale, reliability, long-term availability, and low-carbon generation. Nuclear offers a combination of those characteristics that is difficult to replicate with any single alternative.

This does not mean nuclear will replace renewables, natural gas, battery storage, or the grid.

The more significant development is that hyperscalers are assembling broader energy portfolios—and nuclear is increasingly part of them.

The Market Signal Is Bigger Than Any One Deal

It would be easy to look at recent nuclear announcements as individual corporate sustainability decisions.

That would miss the larger story.

Microsoft's agreement supporting the Crane Clean Energy Center restart in Pennsylvania represents one model: bringing an existing nuclear asset back into operation.

Meta's agreements represent another: extending the operating life of existing plants, increasing their output, and supporting advanced nuclear development.

Amazon is pursuing multiple pathways simultaneously, including existing nuclear generation, new SMRs, and investments in companies developing next-generation reactors.

Google is taking a longer-duration technology approach through Kairos Power, including the first project under an agreement intended to bring advanced nuclear energy onto the grid serving its data center operations.

Different companies.

Different technologies.

Different timelines.

But the strategic objective is remarkably similar.

These companies are not waiting for a single energy technology to solve future demand.

They are creating long-term pathways to firm electricity years before some of that capacity will be needed.

That is the real market signal.

Why Nuclear Fits the Data Center Power Profile

Data centers have unusually demanding energy requirements.

They operate continuously.

Reliability requirements are extremely high.

Large campuses consume significant quantities of electricity around the clock.

And future expansion often requires confidence that additional capacity will remain available over long planning horizons.

Nuclear generation aligns with several of those requirements.

Once operational, nuclear plants can provide high levels of continuous output for extended periods. Unlike weather-dependent resources, their generation is not tied to sunlight or wind conditions.

That makes nuclear particularly interesting as part of a diversified data center energy portfolio.

Renewables can provide large quantities of low-cost clean electricity.

Battery storage can improve flexibility and help balance short-duration variations.

Natural gas can provide dispatchable generation.

Nuclear can contribute long-duration, high-capacity-factor generation.

The strategic opportunity lies in how these resources complement one another.

The emerging data center energy model is not about choosing one winner.

It is about assembling the right mix.

Existing Nuclear May Matter First

Much of the public discussion surrounding nuclear power and data centers focuses on Small Modular Reactors.

SMRs are important.

But the nearer-term nuclear story may be less futuristic.

It may be about the fleet that already exists.

Existing nuclear plants have several characteristics that make them strategically significant.

They already occupy established energy sites.

They already have transmission connections.

They have experienced workforces and operating infrastructure.

And in many cases, they have decades of operational history.

This explains why hyperscaler interest is extending beyond new reactor technologies.

Supporting license extensions, plant restarts, and capacity uprates may provide pathways to additional or preserved firm power without waiting exclusively for an entirely new generation of reactors to reach commercial scale.

The distinction matters.

The nuclear comeback is not one story.

It is at least three:

Preserve what already exists.

Increase output from existing assets.

Develop the next generation of reactors.

Those strategies operate on different timelines, allowing large energy buyers to build a more layered approach to future supply.

Restarts Have Entered the Conversation

Perhaps the clearest symbol of nuclear's renewed relevance is the Crane Clean Energy Center.

Formerly Three Mile Island Unit 1, the reactor was shut down in 2019 for economic reasons.

Microsoft's 20-year power agreement with Constellation helped support plans to restart the unit and return approximately 835 MW of generation to the regional grid.

The significance extends beyond the individual facility.

For years, the energy industry largely discussed nuclear retirements as a one-way process.

Once a plant closed, restarting it was rarely considered a realistic commercial strategy.

The Crane project challenged that assumption.

A large technology buyer provided the long-term commercial commitment needed to help make a restart economically viable.

That introduces a new question for the broader market:

Could long-duration demand from data centers create new economics for existing nuclear assets?

Not every retired plant will be suitable for restart.

Technical condition, licensing, economics, location, grid requirements, and ownership all matter.

But the fact that the model is being pursued at all represents a meaningful change in the energy market.

Uprates Could Unlock Additional Capacity

Another less dramatic but potentially important opportunity is nuclear uprating.

An uprate increases the electrical output of an existing reactor through equipment improvements, engineering modifications, or operational changes.

Compared with constructing an entirely new power plant, the concept is straightforward: generate more electricity from infrastructure that is already operating.

For an energy market searching for additional firm capacity, this deserves attention.

Meta's agreement supporting the Clinton Clean Energy Center in Illinois includes incremental capacity from plant uprates. Constellation has also publicly discussed the potential to add substantial additional output across its nuclear fleet through uprating opportunities.

For data center operators, this is strategically interesting because it illustrates that new energy supply does not always require a new generation site.

Some additional capacity may come from making existing assets more productive.

That principle extends beyond nuclear.

Across the power sector, the next energy buildout will involve both constructing new assets and improving the capabilities of infrastructure already in service.

Hyperscalers Are Creating Long-Term Demand Signals

Nuclear projects operate on long timelines.

That creates one of the industry's biggest commercial challenges.

Developers must commit significant capital years before a facility begins producing electricity.

Long-term customers can change that equation.

A hyperscaler willing to commit to energy purchases over decades provides something extremely valuable:

Demand certainty.

That certainty can support financing, development, equipment procurement, licensing, and supply-chain investment.

This may become one of the most important roles technology companies play in nuclear's next chapter.

They do not need to become nuclear operators.

Instead, they can provide the commercial demand signals that allow experienced utilities, developers, and nuclear companies to invest.

The relationship resembles the role corporate PPAs played in scaling renewable generation.

Large technology companies helped create bankable demand for wind and solar projects.

Now a similar commercial mechanism may help accelerate investment in firm clean generation.

Meta Is Taking the Strategy Further

Meta's nuclear activity provides perhaps the clearest example of how quickly the strategy is expanding.

After its agreement supporting the Clinton Clean Energy Center, Meta announced additional nuclear agreements in 2026 involving Vistra, TerraPower, and Oklo.

Combined with its earlier Constellation agreement, the company says these projects could support up to 6.6 GW of new and existing nuclear energy by 2035.

That is significant not merely because of the number.

It shows that nuclear procurement is moving beyond individual pilot projects.

The strategy spans existing reactors and advanced technologies.

It spans multiple companies.

And it extends across multiple U.S. markets.

This looks less like experimentation and more like portfolio construction.

For the data center industry, that distinction is critical.

Amazon Is Building Multiple Nuclear Pathways

Amazon is pursuing a similarly diversified approach.

Its strategy includes support for existing nuclear capacity at the Susquehanna plant in Pennsylvania, investment in X-energy, collaboration with Energy Northwest on advanced reactors in Washington, and exploration of additional nuclear opportunities.

Amazon and Talen Energy expanded their nuclear relationship in 2025 through an agreement that can supply up to 1,920 MW from the Susquehanna nuclear plant for Amazon operations.

At the same time, Amazon has backed X-energy's advanced reactor technology and a planned Energy Northwest project that could begin with 320 MW and eventually expand substantially.

These two strategies illustrate an important point.

Existing nuclear and advanced nuclear do not compete for the same role.

One helps address near- and medium-term energy needs.

The other creates options for the 2030s and beyond.

For operators planning data center capacity across decades, both timelines matter.

Google Is Testing the Advanced Nuclear Model

Google's partnership with Kairos Power represents another important development.

Rather than beginning with a conventional nuclear plant, Google is helping create a commercial pathway for advanced reactor technology.

The initial agreement was designed around multiple Kairos reactors, with deployments planned into the 2030s.

That strategy advanced further when Google, Kairos Power, and the Tennessee Valley Authority announced a project involving the Hermes 2 plant in Oak Ridge, Tennessee.

The project is expected to provide up to 50 MW to the TVA grid serving Google's data centers in Tennessee and Alabama.

Fifty megawatts is small compared with today's largest hyperscale campuses.

But that misses the strategic significance.

Early advanced reactor projects are about proving technology, regulatory pathways, manufacturing processes, operating models, and commercial structures.

If those pieces work, later deployments can scale.

The first reactor is therefore not simply an energy asset.

It is a test of whether an entirely new nuclear supply chain can become commercially repeatable.

SMRs Still Need to Prove Scale

The enthusiasm around advanced nuclear should be balanced with execution reality.

SMRs promise several potential advantages.

Smaller individual units.

Modular deployment.

Potentially more standardized manufacturing.

Flexible siting.

The ability to add capacity incrementally.

Those characteristics align well with the phased expansion model common to large data center campuses.

But commercial deployment at scale remains ahead of the market.

Cost must be proven.

Construction schedules must be demonstrated.

Fuel supply chains must mature.

Regulatory processes must continue advancing.

Manufacturing capacity must expand.

That is why current hyperscaler investment is important.

The industry is helping fund the period between promising technology and repeatable commercial deployment.

Whether every reactor design succeeds is less important than the broader trend.

Large electricity buyers are willing to invest early because the potential strategic value of firm future power is high enough to justify exploring multiple options.

Nuclear Changes the Timeline of Energy Strategy

Perhaps the biggest implication for data centers is planning horizon.

Data center companies are accustomed to thinking several years ahead.

Nuclear requires thinking much further.

A reactor expected to produce electricity in the early or mid-2030s must be incorporated into energy strategy today.

That changes the nature of procurement.

Instead of asking only:

What power can we secure for the next facility?

Operators increasingly need to ask:

What energy resources should exist ten years from now to support the campuses we have not yet built?

That is an entirely different level of planning.

And it demonstrates how important energy has become to long-term data center strategy.

Nuclear Sites Could Gain Strategic Importance

There is also a geographic implication.

Existing nuclear sites often have characteristics attractive to large energy users.

They are established generation locations.

They frequently have substantial transmission connectivity.

They have experienced energy workforces.

They may have land or surrounding areas capable of supporting additional development.

And local communities already understand large-scale energy operations.

That does not mean data centers will simply cluster around every nuclear facility.

Connectivity, latency, land, water, taxes, workforce, permitting, and utility structure still matter.

But regions with existing nuclear generation or credible pathways to additional nuclear capacity may receive greater attention from large power users.

The power map and the data center map are becoming more closely connected.

The Supply Chain Opportunity Is Larger Than the Reactor

A nuclear expansion cycle would also require investment well beyond reactor sites.

Fuel.

Specialized components.

Engineering.

Construction.

Electrical equipment.

Control systems.

Operations.

Maintenance.

Workforce development.

Advanced reactor manufacturing.

The supply chain implications are substantial.

This matters because one of the biggest challenges facing America's broader power buildout is not simply identifying technologies.

It is building enough equipment, labor capacity, and project expertise to deploy those technologies repeatedly.

Corporate nuclear agreements create demand signals throughout that ecosystem.

A commitment to future generation can encourage suppliers to invest today.

That multiplier effect may become one of the most important consequences of hyperscaler involvement.

Nuclear Is Not a Standalone Strategy

The strongest interpretation of the nuclear comeback is not that the industry has discovered one final answer to data center energy.

It is the opposite.

The energy strategy is broadening.

Large operators are increasingly combining renewable procurement, utility power, storage, natural gas, geothermal, demand flexibility, grid investment, and nuclear energy.

Each resource solves a different part of the equation.

Nuclear's role is particularly valuable where operators need long-duration, high-output, low-carbon electricity.

Storage provides flexibility.

Renewables provide scalable clean generation.

Natural gas can provide dispatchable capacity.

Grid investment connects resources with demand.

Demand response can improve how existing capacity is used.

The emerging model is diversified by design.

That is why nuclear's return should not be viewed as a rejection of previous energy strategies.

It is another layer being added to them.

The Corporate Energy Buyer Is Changing

Perhaps the most important long-term implication has little to do with nuclear technology itself.

The role of the corporate energy buyer is changing.

Historically, a data center operator primarily purchased electricity.

Then major technology companies became some of the world's largest renewable energy buyers.

Now they are helping preserve power plants, support restarts, fund reactor developers, commit to future generation, and influence which energy technologies reach commercial scale.

That represents a profound expansion of responsibility.

Energy procurement is no longer simply about finding supply.

For the largest operators, it increasingly involves helping create the future supply they expect to need.

That may ultimately be the biggest lesson from nuclear's comeback.

What It Means for the Broader Data Center Market

Most data center operators will never sign a multi-gigawatt nuclear agreement.

That does not mean the trend is irrelevant to them.

Hyperscaler investments can influence regional generation supply, transmission investment, nuclear supply chains, technology costs, utility planning, and future availability of firm power.

If advanced nuclear reaches commercial scale, its customers will eventually extend beyond the handful of companies funding the earliest projects.

If existing nuclear plants remain online longer, the regions they serve retain valuable generation capacity.

If uprates add new output, more electricity becomes available to the broader system.

The benefits can extend beyond the original corporate buyer.

That makes nuclear development an industry-wide energy story, not simply a hyperscaler story.

The Next Five Years Matter

The nuclear comeback has momentum.

But the next several years will determine whether that momentum translates into durable capacity.

The Crane restart must progress through its remaining regulatory and engineering milestones.

Existing plants must secure long-term operating certainty.

Uprate projects must demonstrate their economics.

Advanced reactor developers must move from demonstration projects toward repeatable deployment.

Supply chains must expand.

And utilities, regulators, technology companies, and nuclear developers must continue building workable commercial structures.

There is substantial opportunity.

There is also substantial execution ahead.

That is precisely why the current moment matters.

The contracts being signed today will help determine what America's power system can deliver in the 2030s.

Nuclear Has Reentered the Energy Portfolio

The data center industry's relationship with nuclear energy has changed dramatically.

What once appeared to be a specialized or long-term option is now part of active energy strategy at some of the world's largest technology companies.

Existing plants are receiving long-term support.

Retired generation is being considered for restart.

Uprates are creating additional capacity.

Advanced reactor developers are receiving corporate investment.

New commercial models are being tested.

This does not mean every data center will be powered by nuclear energy.

Nor does it mean nuclear will replace the other resources supporting America's growing electricity needs.

The more important development is that nuclear has earned a seat at the table.

As data centers plan for larger campuses and longer energy horizons, the industry is looking for every credible source of reliable, scalable power.

Nuclear offers one of those pathways.

And if the commitments from Microsoft, Google, Amazon, Meta, and others are any indication, the nuclear comeback is no longer just an energy-sector story.

It is becoming a data center energy story.