Introduction: Preparing the Ground Before the Reactor — or the AI Factory — Appears
One of the more consequential nuclear-energy developments of 2026 did not begin with a reactor vessel, a cooling tower, a hyperscale data center, or a multibillion-dollar construction announcement. It began with something much less dramatic: a state government creating a way for ordinary communities to say, in advance, we are preparing ourselves for nuclear energy.
In July 2026, Indiana Governor Mike Braun announced the launch of the Indiana Advanced Nuclear Ready Community program, administered by the state’s Office of Energy Development and described by the state as a first-of-its-kind designation system. Cities, counties, and townships can now work toward a set of escalating tiers of formal recognition by organizing local stakeholders, conducting public education, screening sites, developing workforce capacity, establishing relationships with reactor developers, and taking the other unglamorous steps that make a future nuclear project easier to execute. The program is explicitly connected to Indiana’s changing electricity system, its position along the PJM–MISO seam, and the accelerating electricity demand associated with data centers.[1][3]
The governor framed the program in the language not of technology but of household economics:
“Affordability is why advanced nuclear energy is a top priority.” — Governor Mike Braun [2]
His Secretary of Energy and Natural Resources, Suzanne Jaworowski, framed it in the language of risk and time:
“Indiana is significantly reducing developer risk and accelerating timelines for advanced energy investment.” — Suzanne Jaworowski, Indiana Secretary of Energy and Natural Resources [52]
That may sound bureaucratic compared with the spectacle of a new reactor rising out of the ground. But it represents an important change in how the nuclear economy is being organized — and, this paper argues, an early glimpse of how the geography of the American AI economy will be decided.
Traditionally, the sequence of nuclear development ran in one direction. A reactor developer identified a technology and a site. A utility or industrial customer identified a need for power. Regulators began licensing. And the surrounding community then confronted the consequences of a project that had already become tangible — often as the last party to the table rather than the first. In the emerging AI economy, states are beginning to reverse that sequence. They are trying to prepare communities, infrastructure, regulatory institutions, workforces, supply chains, and political coalitions before the ultimate electricity customer has even been identified, let alone finalized.
The timing of Indiana’s move was not accidental, and the week during which this paper was completed illustrates why.
During the week of August 13–14, 2026 alone, two developments crystallized the moment. First, X-energy disclosed on its second-quarter earnings call that the U.S. Department of Energy had formally notified the company that its cooperative agreement under the Advanced Reactor Demonstration Program would receive up to an additional $1 billion — bringing total federal cost-shared support for its Texas project with Dow to as much as $2.15 billion since 2021.[4] X-energy’s chief executive, Clay Sell, told investors:
“Our ARDP cooperative agreement will receive up to an additional $1 billion.” — Clay Sell, CEO, X-energy [5]
Second, TerraPower and South Korea’s SK Innovation signed a preliminary global agreement on small modular reactor projects at a ceremony in Seoul attended by TerraPower founder Bill Gates, SK Group Chairman Chey Tae-won, and Korea’s industry minister — an agreement that opens the door for one of Korea’s largest industrial groups to participate in TerraPower’s Natrium projects in the United States and abroad, and that arrived alongside a component-manufacturing contract for Doosan Enerbility covering the Natrium reactor vessel and internal structures.[6][7] The presence of Gates in Seoul — meeting in a single day with SK, HD Hyundai, and senior Korean officials to discuss power supply for artificial intelligence — was itself a signal of how deeply advanced nuclear has moved into the strategic calculations of technology investors, industrial conglomerates, governments, and international supply chains.[48]
The larger electricity numbers explain the urgency behind all of this activity. The U.S. Energy Information Administration’s August 2026 Short-Term Energy Outlook projects that American electricity consumption, having set its second consecutive annual record in 2025 at roughly 4,195 billion kilowatt-hours, will establish new records in both 2026 and 2027, with AI-oriented data centers among the principal drivers — and with commercial electricity demand expected to outpace residential demand for the first time on record.[8] The EIA’s Annual Energy Outlook 2026 estimates that data-center servers alone already accounted for approximately 7 percent of commercial-sector electricity consumption in 2025, a share projected to rise to between 22 and 33 percent of commercial building electricity use by 2050.[9] The Department of Energy, for its part, now describes the national power system as moving from decades of relatively stagnant demand toward unusually rapid load growth driven substantially by hyperscale AI infrastructure — a framing embedded in everything from its transmission planning studies to its nuclear financing programs.[25]
The most quotable summary of the situation belongs to the director of MIT’s Center for Advanced Nuclear Energy Systems, Jacopo Buongiorno, who quantified the American gap in a widely circulated interview:
“Electricity demand will grow by approximately 50 gigawatts just to support data centers and AI.” — Jacopo Buongiorno, MIT [39]
Fifty gigawatts, for perspective, exceeds the installed generating capacity of most European nations. And the head of the International Energy Agency, Fatih Birol, has reduced the entire dynamic to a single sentence delivered to world leaders at the AI Action Summit in Paris:
“There is no AI without energy — specifically electricity.” — Fatih Birol, Executive Director, International Energy Agency [41]
The result of all this is a change in the geography of artificial intelligence.
For much of the cloud era, technology companies could decide where to locate computing infrastructure primarily by comparing fiber connectivity, land prices, taxes, skilled labor, latency, and proximity to customers. Electricity was obviously important, but electricity availability rarely dominated corporate strategy the way it increasingly does for gigawatt-scale AI campuses. In the emerging Five-Layer AI Economy, that assumption no longer holds. The first layer — Energy — increasingly determines how quickly the remaining layers can expand.
| Layer | Function | What Happens When It Binds |
| Layer 1: Energy | Supplies the electricity | Every layer above it waits |
| Layer 2: Chips | Turns electricity into accelerated computation | GPUs sit dark, depreciating |
| Layer 3: Data Centers | Aggregates chips into physical AI factories | Buildings complete but unenergized |
| Layer 4: Models | Converts computation into machine intelligence | Training runs delayed or relocated |
| Layer 5: Applications & Agents | Distributes intelligence through the economy | Product roadmaps slip; revenue deferred |
Table 1. The Five-Layer AI Economy and the consequences of an energy-layer constraint.
When Layer 1 becomes constrained, additional GPUs do not automatically produce additional intelligence. A warehouse containing hundreds of thousands of advanced accelerators can become economically stranded if the utility cannot provide the megawatts required to energize it. The scale of capital exposed to this risk is extraordinary: through the second-quarter 2026 earnings season, the four largest hyperscalers — Amazon, Microsoft, Alphabet, and Meta — had guided to roughly $725 billion in combined capital expenditure for 2026 alone, up approximately 77 percent from an already record 2025, with the overwhelming majority directed at AI data centers, accelerators, custom silicon, and power.[43][44]
That is why nuclear power has reentered the AI discussion with such force.
Meta announced agreements in January 2026 involving Vistra, TerraPower, and Oklo that, together with its earlier Constellation contract, could support as much as 6.6 gigawatts of existing and future nuclear capacity by 2035 — all of it destined for the PJM region, and explicitly connected by the company to the electricity requirements of its AI infrastructure, including the Prometheus supercluster in Ohio.[10][11][12] Amazon has backed X-energy, whose April 2026 initial public offering raised more than $1 billion and left the company with $1.9 billion in cash and no debt.[49] Google has supported advanced-reactor development through Kairos Power. Microsoft’s twenty-year power purchase agreement underpins Constellation’s restart of the former Three Mile Island Unit 1 in Pennsylvania, now the Christopher M. Crane Clean Energy Center.[13]
But these announcements expose a difficult mismatch.
AI infrastructure moves quickly. A technology company can order GPUs, raise capital, sign land agreements, and begin developing a data-center campus within a relatively short investment cycle. Nuclear infrastructure does not move on the same clock. Nuclear projects require licensing, engineering, fuel, specialized components, transmission connections, emergency planning, trained operators, construction workforces, security arrangements, public participation, political durability, and very large amounts of patient capital. Even the restart of an existing nuclear station involves thousands of inspections, maintenance activities, regulatory actions, and grid questions. Palisades in Michigan illustrates the point: although Holtec announced in July 2026 that the restart project had reached a “watershed moment” with the last major projects closed out, more than 5,000 individual work activities still remained on the checklist, and the NRC continued its oversight of the first-of-a-kind return of a permanently shutdown plant to operations.[16][17][18]
This mismatch creates the central question of this paper:
What if states stop waiting for a hyperscaler to arrive before preparing the nuclear-energy ecosystem that hyperscalers will eventually need?
That question is why I chose the title Reactor Readiness. The term deliberately shifts attention away from the reactor itself. The defining competitive advantage of the next decade may not be possession of a particular reactor design. Many states may eventually have access to broadly similar technologies from companies such as TerraPower, X-energy, Kairos Power, Oklo, Holtec, Westinghouse, GE Vernova Hitachi, or the emerging generation of microreactor developers — Antares, Valar Atomics, Aalo, Radiant, and others whose test reactors reached criticality under the Department of Energy’s Reactor Pilot Program in the summer of 2026.[20][21]
What states will not possess equally is readiness.
A reactor design can theoretically be sold in many jurisdictions. A community that already understands nuclear development, controls an appropriate site, possesses transmission access, has water and infrastructure plans, has organized labor and educational institutions, understands emergency planning, has established state permitting coordination, enjoys durable political support, and has begun conversations with potential industrial electricity buyers provides something much scarcer. It provides time certainty. And in an AI economy where billions of dollars of GPUs depreciate rapidly while companies race to bring new computing capacity online, time certainty can become almost as valuable as the electricity itself.
I therefore define Reactor Readiness as:
The institutional, physical, regulatory, economic, workforce, and civic capacity of a jurisdiction to move an advanced nuclear project from serious commercial interest toward deployment without first spending years constructing the surrounding ecosystem from zero.
This definition separates Reactor Readiness from several related concepts, and the distinctions matter for policy design.
It is not simply nuclear advocacy. A community can support nuclear energy rhetorically without possessing an appropriate site, workforce, grid connection, permitting system, or developer relationship. Rhetoric is free; readiness is built.
It is not reactor construction. Readiness occurs largely before construction, in the years when a project’s ultimate cost and schedule are quietly being determined by decisions about land, transmission, training, and process.
It is not simply a data-center incentive. The nuclear asset may ultimately serve industrial manufacturing, households, defense installations, semiconductor fabs, AI facilities, or several categories simultaneously — and, as the X-energy/Dow project demonstrates, the first-of-a-kind learning may be financed by a chemical plant rather than a cloud provider.[4]
And it is not merely an energy policy. Reactor Readiness sits at the intersection of energy policy, industrial policy, AI policy, land-use policy, workforce development, infrastructure finance, local government, and national security.
The central argument of this paper is that advanced nuclear competition is becoming a competition among prepared places. The states and communities that understand this early may acquire a powerful and compounding advantage in the next phase of America’s AI infrastructure buildout.
The paper proceeds in six sections. Section 1 examines how the AI power shock is converting nuclear readiness from an energy question into economic-development policy. Section 2 surveys the emerging laboratory of state-level readiness strategies — Indiana, Texas, Michigan, Tennessee, Virginia, and Pennsylvania — and extracts the distinct model each represents. Section 3 develops a five-dimension Reactor Readiness Framework and a 100-point Reactor Readiness Index. Section 4 analyzes the transformation of hyperscalers into nuclear market makers. Section 5 offers a ten-step policy roadmap for states for 2026–2035. Section 6 distills the argument into seven pillars, and the conclusion returns to the deeper meaning of readiness in an economy whose binding constraint has become the electron.
Section 1: The AI Power Shock Is Turning Nuclear Readiness into Economic Development Policy
1.1 From Electricity Consumption to Electricity Strategy
Artificial intelligence has transformed electricity from an operating expense into a strategic input, and this transformation is the foundation on which every argument in this paper rests. It is therefore worth dwelling on what, precisely, has changed.
A traditional commercial building consumes electricity without fundamentally changing the utility system around it. Its load is small relative to the regional system, predictable in shape, and served by infrastructure that was planned decades in advance for exactly this kind of incremental growth. A hyperscale AI campus is categorically different. Proposed campuses now routinely involve hundreds of megawatts, and the most ambitious development plans span multiple gigawatts. At that scale, a single corporate siting decision affects transmission planning, generation construction, wholesale market prices, regulatory policy, local politics, and sometimes statewide economic-development strategy. The electricity system no longer sits quietly beneath the digital economy. It increasingly determines its geography.
The macro data confirm the shift. The EIA’s Short-Term Energy Outlook now projects U.S. power demand rising from a record of roughly 4,195 billion kilowatt-hours in 2025 to approximately 4,268 billion kilowatt-hours in 2026 and roughly 4,391 billion kilowatt-hours in 2027 — the third and fourth consecutive annual records — driven in large part by data centers dedicated to artificial intelligence.[8] Notably, the same August 2026 outlook contained a warning about how quickly this geography can shift: after the Texas governor announced a pause on new data-center development on August 3, 2026, to collect more information about projects under review, the EIA lowered its Texas load-growth forecast for 2027 from 14 percent to 6 percent.[8] A single state-level policy decision moved a national energy forecast. That is what it means for electricity to have become strategic.
The International Energy Agency has framed the same phenomenon globally. In releasing its landmark Energy and AI report, Executive Director Fatih Birol observed that data centers are on course to account for almost half of the growth in U.S. electricity demand, and offered a comparison designed to be remembered:
“Consuming as much electricity by 2030 as the whole of Japan does today.” — Fatih Birol, on global data-center demand [42]
That transition creates two races occurring simultaneously. Technology companies are racing to obtain compute capacity. States are racing to obtain generation capacity. The two races eventually converge on the same substations, the same transmission corridors, and the same interconnection queues. A state may offer cheap land, tax incentives, fiber routes, research universities, and political support, but if a proposed AI campus cannot obtain several hundred megawatts within its required schedule, those advantages become secondary. This is where nuclear power becomes more than an environmental policy preference. It becomes economic-development infrastructure.
1.2 Why Nuclear Has Returned to the AI Conversation
Nuclear electricity possesses a bundle of characteristics unusually compatible with large computational facilities, and it is the bundle — not any single attribute — that explains its return to the center of the conversation.
It operates at high capacity factors, typically above 90 percent across the U.S. fleet, which matters to facilities designed to run continuously. It produces very large quantities of electricity from relatively compact sites, which matters where land near load is scarce. It provides firm generation rather than depending primarily on weather, which matters to customers whose service-level commitments do not pause for a windless week. Its fuel occupies little physical space and can be stored on site for years, which matters for energy security. Plants operate for four, six, or potentially eight decades, which matters to companies making thirty-year infrastructure commitments. Existing nuclear sites often possess transmission infrastructure, skilled workforces, security perimeters, community familiarity, and industrial land — a preassembled bundle of exactly the assets that greenfield projects spend years acquiring. And advanced reactor developers argue that smaller designs may eventually provide greater siting flexibility and factory-driven repeatability, although that proposition remains to be proven at commercial scale.
None of these characteristics means nuclear is automatically the cheapest or fastest solution. Natural-gas generation, renewables paired with storage, grid imports, geothermal, and demand flexibility will all play important roles, and in many places larger near-term roles. Princeton’s Jesse Jenkins and collaborators have shown, for example, that even modest data-center load flexibility — the willingness to curtail a small fraction of hours per year — can unlock dramatically faster interconnection and defer tens of millions of dollars in grid upgrades, a finding that argues against treating any single supply technology as a silver bullet.[40]
The stronger proposition, and the one this paper adopts, is that AI electricity demand is now large enough that nuclear no longer needs to defeat every alternative. It merely needs to become economically valuable within an increasingly constrained portfolio. That distinction is important. The AI economy may require addition rather than substitution. A state adding ten gigawatts of new computational load may need renewables, natural gas, nuclear generation, transmission expansion, storage, efficiency, and flexible-load arrangements simultaneously. The nuclear question therefore becomes less ideological — should nuclear replace another technology? — and more practical: how many additional reliable megawatts can the state create, how quickly, and where?
1.3 The Technology Clock and the Reactor Clock
The fundamental problem is that the AI investment cycle and the nuclear-development cycle operate at different speeds, and the gap between them is where fortunes will be made and lost.
A hyperscaler makes major capital-allocation decisions on a three-to-five-year infrastructure horizon, under competitive pressure measured in months. The hardware inside an AI data center depreciates on schedules of roughly four to six years, and the frontier models it trains can be commercially superseded within one. A first-of-a-kind advanced reactor, by contrast, can require a decade or more from concept to commercial operation, and even “fast” projects — restarts of existing plants, uprates, early-site-permitted SMRs — operate on multi-year schedules governed by licensing, procurement, and construction realities that no earnings call can compress.
That mismatch is precisely why readiness becomes valuable. If state and local governments wait until a hyperscaler requests power before beginning site identification, zoning, environmental analysis, transmission studies, workforce preparation, public engagement, nuclear supply-chain development, local emergency planning, state permitting coordination, and developer outreach, the power solution may arrive too late for the customer’s original investment schedule — and the customer, along with its capital, will go somewhere that prepared earlier.
The alternative is to perform portions of that work speculatively. States, in effect, create an inventory of future nuclear options. This is the same logic by which economic-development agencies have long prepared certified industrial parks before any manufacturer has committed to build a factory: roads are mapped, utilities are evaluated, land is assembled, workforce programs are created, and local officials learn the permitting process in advance, so that when the investor arrives, the answer to “how long?” is measured in months rather than years. Nuclear development is beginning to adopt the same logic, with Indiana’s community-designation program as the clearest early expression.[1][2]
1.4 The Federal Government Is Shortening One Side of the Equation
The federal government is simultaneously attempting to accelerate the reactor side of the equation, and the record of the twelve months preceding this paper is genuinely remarkable — a burst of demonstrations, awards, and financing commitments with few precedents in the modern history of the industry.
The Department of Energy’s Reactor Pilot Program, created under a May 2025 executive order directing DOE to achieve criticality for at least three advanced reactors by July 4, 2026, selected eleven projects from ten companies and then beat its own deadline.[20] On June 4, 2026, Antares Nuclear’s Mark-0 microreactor achieved initial criticality at Idaho National Laboratory — the first privately developed advanced reactor to do so under the program. Energy Secretary Chris Wright called it:
“A historic moment for American nuclear energy.” — Chris Wright, U.S. Secretary of Energy [22]
Two weeks later, Valar Atomics’ Ward 250 high-temperature gas reactor achieved zero-power criticality at Utah’s San Rafael Energy Lab — the first DOE-authorized reactor built and operated outside the national laboratory system, and a machine that had been airlifted between Air Force bases aboard a C-17 before being fueled on site. It generated thermal power on June 21 and produced electricity on July 1.[21][80] By Independence Day, a third and then a fourth criticality had followed, including Aalo Atomics’ sodium-cooled Critical Test Reactor at INL.[46][50] INL Director John Wagner captured the institutional meaning:
“American nuclear innovation has its momentum back.” — John Wagner, Director, Idaho National Laboratory [21]
The demonstrations were accompanied by deployment money. In May 2026, DOE awarded more than $94 million to eight companies under the Generation III+ SMR Pathway to Deployment Program — the second tranche of a roughly $900 million initiative whose first $800 million had gone to the Tennessee Valley Authority and Holtec — funding early site permits in New York and Nebraska, reactor-pressure-vessel assembly capacity in Indiana, forging furnaces in Pennsylvania, fuel-fabrication lines, and nuclear quality-assurance certifications across eight states.[23] Secretary Wright framed the purpose in explicitly AI-inflected terms:
“Reliable, round-the-clock power we need to fuel the President’s manufacturing boom.” — Chris Wright, on Gen III+ SMR awards [24]
Then, in June 2026, came the largest single action: DOE’s Office of Energy Dominance Financing issued a conditional commitment for $17.5 billion in American Nuclear Supply Chain Loans to finance long-lead components — reactor pressure vessels, steam generators, reactor coolant pumps, prefabricated structural modules — for up to ten Westinghouse AP1000 reactors at five U.S. sites, with the explicit aim of accelerating deployment by up to three years and of meeting the federal objective of ten large reactors under construction by 2030.[25][26] The structure is instructive: Westinghouse and each utility partner must commit $500 million in equity apiece — $1 billion per project — before accessing federal financing, and DOE officials indicated the projects would be supported by power purchase agreements of up to twenty-five years with hyperscalers, priced to ensure the projects proceed without raising electricity costs in host regions.[26] Wright described the goal as:
“Reviving the supply chain needed for America to once again build large-scale commercial reactors.” — Chris Wright, on the $17.5 billion supply-chain loans [25]
In August 2026, the additional $1 billion for the X-energy/Dow Seadrift project — expected to be the first grid-scale advanced reactor serving an industrial site in North America — extended the same pattern to Generation IV technology.[4][49]
These actions matter for this paper because federal reactor acceleration and state community readiness are complementary, not substitutes. Washington can help mature reactor technology, de-risk first-of-a-kind supply chains, and mobilize capital at a scale no state can match. But Washington cannot single-handedly manufacture local political acceptance, county-level zoning, regional workforce programs, transmission access, municipal infrastructure, or community trust in every potential host location. Those assets are produced locally or not at all. That is where states become the decisive actors — and where the concept of Reactor Readiness does its work.
1.5 Reactor Readiness Begins Before Reactor Selection
An important design feature of the concept is technological neutrality. A reactor-ready jurisdiction should not necessarily begin by deciding that only one reactor technology will be acceptable. Its first task is understanding the characteristics of its own sites and markets, because the mapping between technologies and applications is many-to-many and still evolving.
A gigawatt-scale industrial region with strong transmission may eventually support conventional large reactors — the AP1000 class now backed by the federal supply-chain loans. An isolated military facility might benefit from a transportable microreactor of the kind Antares and Valar are demonstrating. A retired coal site might suit a 300-megawatt-class SMR, as TerraPower’s coal-community project in Kemmerer, Wyoming, is designed to prove. An existing nuclear campus may accommodate additional units, as Holtec proposes with its SMR-300 “Pioneer” pair at Palisades.[19] A chemical facility may require both steam and electricity, the configuration X-energy is building for Dow at Seadrift. An AI campus may prioritize firm electricity, expansion optionality, and a predictable energization schedule above all else.
Readiness therefore begins with capabilities and constraints, not with a favored vendor. This ordering prevents states from becoming prematurely locked into technologies whose commercialization schedules may slip — a real risk in a field where every first-of-a-kind project carries schedule uncertainty. The goal of a readiness program should be to create places where several qualified developers could plausibly compete. Competition among prepared developers for prepared places is the market structure most likely to discipline cost, and it is only possible if the places prepare first.
Section 2: America Is Becoming a Laboratory of State-Level Reactor Readiness
If the first section established why readiness matters, this section establishes that it is already happening — unevenly, experimentally, and through instruments so different from one another that the United States now functions as a natural laboratory of readiness strategies. Six states anchor the analysis. Each has chosen a different lever, and each teaches a different lesson.
| State | Primary Instrument | Core Mechanism | What It Teaches |
| Indiana | Advanced Nuclear Ready Community program (2026) | Tiered community designation; toolkit; site & workforce preparation | Community preparedness can be standardized and credentialed |
| Texas | TANEO + $350M Advanced Nuclear Development Fund (2025–26) | State coordination office; construction & supply-chain reimbursement | A state can build a nuclear industry, not just buy nuclear power |
| Michigan | Palisades restart + SMR-300 siting + ratepayer plan | Asset recovery; site reuse; data-center cost responsibility | Existing nuclear campuses are strategic inventory |
| Tennessee | Nuclear Energy Fund ($70M) + Oak Ridge cluster | Fuel-cycle, manufacturing, and workforce investment | Readiness has a supply-chain geography, not just a host geography |
| Virginia | Data-center consumption tax + SCC cost allocation + Power Innovation | Large-load cost responsibility paired with nuclear innovation | Demand-side discipline and supply-side ambition can coexist |
| Pennsylvania | GRID Standards + Crane restart ecosystem | Conditioning data-center benefits on energy & community commitments | The grid, not the reactor, can be the critical path |
Table 2. Six state models of Reactor Readiness, 2026.
2.1 Indiana: Turning Nuclear Preparation into a Community Credential
Indiana provides perhaps the clearest and most transferable expression of the Reactor Readiness idea, precisely because it operates at the level most nuclear policy ignores: the county commission, the township board, the local school district.
The Advanced Nuclear Ready Community program establishes a structured, three-tier designation framework through which communities progress by demonstrating increasing levels of preparation — from public education meetings that build local understanding of nuclear energy, through stakeholder organization and site screening, to workforce partnerships and formal engagement with developers.[1][3] The state’s Office of Energy Development supports the process with a digital toolkit of templates and reference materials, so that a county of forty thousand people does not need to reinvent the machinery of nuclear-hosting preparation from first principles.[52] The program’s stated aims are explicit about the economics: ensure community buy-in, reduce developer risk by completing upfront site assessments before developers engage, attract manufacturers by demonstrating regulatory familiarity, and build local workforce pipelines with educational institutions.[2]
This matters because Indiana is treating nuclear readiness as a recognizable, portable economic-development asset. A community that earns a designation can effectively signal to reactor developers, utilities, investors, manufacturers, and future electricity buyers: you will not need to begin this conversation with us from zero. That signaling function could prove powerful. Hyperscalers conduct extensive site-selection exercises; reactor developers do the same; infrastructure investors evaluate execution risk; supply-chain companies search for geographic clusters. A credible readiness designation reduces information costs for all of them simultaneously — which is exactly what certification systems are for. LEED did this for buildings. Certified-sites programs did it for industrial parks. Indiana is attempting to do it for the nuclear host community.
Indiana’s program may therefore represent the early version of something that spreads nationally: standardized nuclear-host-community preparation, potentially with reciprocity across states, federal recognition, or integration into DOE siting initiatives. The program’s location matters too. Indiana sits astride the PJM–MISO seam, where firm baseload capacity has unusual option value, and the state has framed the program candidly as part of its response to accelerating data-center demand.[1] The first designations will be worth watching closely: the moment a designated Indiana community lands a project that a non-designated peer community loses, the program will have proven its market logic, and imitation elsewhere will follow quickly.
2.2 Texas: From Nuclear Interest to Nuclear Industrial Policy
Texas is approaching the problem from the opposite end of the scale — not the township but the state-industrial complex.
In June 2025, Governor Greg Abbott signed House Bill 14, establishing the Texas Advanced Nuclear Energy Office (TANEO) and creating a $350 million Texas Advanced Nuclear Development Fund.[27] TANEO provides strategic leadership for the advanced nuclear industry and its supply chain, serves as a single point of contact for developers navigating state-level permitting, and administers the fund through two reimbursement programs: an Advanced Nuclear Construction Reimbursement Program covering NRC licensing costs, long-lead component procurement, and construction activities; and a Project Design and Supply Chain Reimbursement Program covering design, manufacturing capacity, and supply-chain development.[27][28] Applications for the full $350 million opened on April 1, 2026, with awards selected in the summer — a pace that itself signals intent.[28] Abbott has been characteristically direct about the ambition:
“We will jumpstart next-generation nuclear development and deployment.” — Governor Greg Abbott [29]
The strategically significant point is that Texas is not merely attempting to purchase nuclear electricity. It is trying to create a nuclear industry — developers, licensing capability, manufacturing capacity, fuel-cycle activities, and construction expertise, resident in-state and exportable to projects elsewhere. Texas already possesses many of the complementary assets that matter for AI infrastructure: the nation’s largest competitive energy market, enormous generation resources, deep industrial and petrochemical expertise, semiconductor manufacturing, available land, and one of the fastest data-center buildouts in the world. The X-energy/Dow Seadrift project — now backed by up to $2.15 billion in federal cost share — gives the state a flagship Generation IV industrial deployment, while the governor’s August 2026 pause on new data-center development, whatever its ultimate duration, demonstrated that even Texas insists on sequencing growth against infrastructure.[4][8]
Texas therefore illustrates how Reactor Readiness can expand beyond host-community preparation to encompass the full stack: developer readiness + supply-chain readiness + capital readiness + regulatory readiness. The TANEO model — a nonregulatory, single-point-of-contact office that coordinates without displacing substantive safety review — deserves particular attention from other states, and this paper returns to it in Section 5.
2.3 Michigan: Restarting the Past While Preparing the Next Generation
Michigan provides a third model, built not on new programs but on the recovery of an existing asset — and it may be the most instructive case in the country about what readiness actually costs when it has been allowed to lapse.
The Palisades Nuclear Plant on Lake Michigan stopped operating in May 2022, a casualty of merchant-market economics. Holtec International purchased it for decommissioning, then reversed course as the demand environment transformed, pursuing the first restart of a permanently shutdown U.S. commercial nuclear plant with the support of a $1.5 billion DOE loan facility and state funding. The regulatory and physical undertaking has been enormous. In July 2025 the NRC approved the licensing package returning the plant to operations status; through 2025 and 2026 the project moved through steam-generator repairs, turbine-generator refurbishment, fuel receipt, emergency-planning exercises, and thousands of discrete restart activities under a dedicated NRC inspection regime.[18] In July 2026, Holtec announced that the last major projects had closed out and the site had transitioned to final testing, inspection, and operational readiness — the “watershed moment” — while acknowledging that more than 5,000 individual work activities remained and declining to commit publicly to a firm restart date.[16][17] Founder and CEO Kris Singh told the Financial Times he expects the plant to:
“Restart this year, ahead of its contract to supply power by March 2027.” — Kris Singh, CEO, Holtec International [17]
The Palisades story becomes even more relevant because the site will not stop with the existing 800-megawatt reactor. In January 2026, Holtec submitted the first part of a construction permit application to the NRC for two SMR-300 units — Pioneer 1 and 2 — at the same location, including a request for limited work authorization to begin certain construction activities early.[19] Michigan therefore illustrates a powerful compounding pathway: existing nuclear site → reactor restart → retained workforce → preserved transmission → additional advanced reactors → regional energy anchor. Every element of that chain is a readiness asset that most states would need a decade to assemble and that Michigan nearly allowed to dissipate in 2022.
Governor Gretchen Whitmer’s administration has supported the restart while simultaneously moving, in July 2026, to ensure that the data-center demand now flooding into Michigan pays its own way. Her Michigan Affordable and Responsible Growth Action Plan pairs proposed legislation codifying the Public Service Commission’s large-load safeguards with a voluntary pledge under which data-center companies commit to cover their own energy, water, and infrastructure costs.[30] Whitmer’s framing was unambiguous:
“Working families do not pay a single penny for data center development.” — Governor Gretchen Whitmer [31]
That combination is the real Michigan lesson. A pro-AI energy strategy does not require states to subsidize unlimited electricity demand. A more durable model is emerging: build more firm generation while simultaneously requiring large loads to pay their own way. As electricity prices become a first-order political issue heading into the November 2026 elections, states that cannot demonstrate both halves of that bargain may find their nuclear ambitions collapsing under affordability politics.
2.4 Tennessee: Building the Supply Chain Before the Fleet
Tennessee represents a fourth and structurally different form of readiness — one measured not in host communities but in centrifuges, forgings, fuel lines, and engineers.
Governor Bill Lee’s Nuclear Energy Fund, created in 2023 at $50 million and expanded in subsequent budgets to $70 million, has now supported eight companies across nuclear manufacturing, fuel-cycle, and site-development activities.[32] The January 2026 Centrus announcement is emblematic: the company will transition its Oak Ridge facility — the only uranium-enrichment centrifuge manufacturing plant in the United States — to high-rate production, investing more than $560 million and creating nearly 430 jobs to build the thousands of advanced centrifuges that will expand enrichment capacity in Ohio beginning in 2029.[32] Governor Lee’s summary doubles as the state’s thesis:
“Tennessee is the nation’s epicenter for new nuclear.” — Governor Bill Lee [32]
The claim has an increasingly serious evidentiary base. The roster of announced nuclear investments in Oak Ridge alone now includes Orano USA’s multibillion-dollar centrifuge enrichment plant, Oklo’s $1.68 billion nuclear-fuel recycling and advanced fuel center, a $1.4 billion laser-enrichment facility, X-energy’s TRISO-X fuel plant, Radiant’s microreactor manufacturing facility, TVA’s Clinch River SMR project — the first Tier 1 recipient under DOE’s Gen III+ program — and a fusion prototype, layered atop Oak Ridge National Laboratory, TVA, research universities, and a deep energy workforce.[33] Tennessee has even proposed itself to DOE as host of a “nuclear lifecycle innovation campus” spanning enrichment through used-fuel management.[33]
This demonstrates something important about the geography of readiness: Reactor Readiness is not only about communities adjacent to future plants. A state can become reactor-ready by becoming indispensable upstream. Reactors deployed in Indiana, Texas, Virginia, Wyoming, or Ontario may depend on fuel, components, engineers, testing, and manufacturing originating in Tennessee. Readiness therefore has both a host geography and a supply-chain geography, and the supply-chain geography may ultimately capture more durable economic value — just as the semiconductor industry’s profits concentrate in equipment and materials clusters far from many of the fabs themselves.
2.5 Virginia: Converting Energy Pressure into Nuclear Site Preparation and Cost Discipline
Virginia offers the strongest direct link between data-center growth and energy policy, because Virginia is where the collision arrived first and hardest.
The Commonwealth hosts the world’s most important concentration of data-center infrastructure in Northern Virginia — the largest single driver of load growth in the PJM Interconnection — and the resulting pressures on transmission expansion, generation adequacy, consumer costs, land, and water have made data-center policy the central energy question of Virginia politics. Governor Abigail Spanberger’s response in 2026 has been a study in pairing accommodation with discipline. On June 30, 2026, she signed a biennial budget that preserved the data-center sales-tax exemption on equipment while creating the first statewide electricity-consumption tax on data centers in the United States: $0.011 per kilowatt-hour on qualifying facilities, applied whether the power comes from a utility, a competitive supplier, or on-site generation.[34][35] She explained the principle plainly:
“We want data centers to pay their fair share.” — Governor Abigail Spanberger [34]
The budget was followed within days by an energy-affordability package — backup-generator emissions standards, new local assessment tools, ratepayer protections — and then, in August 2026, by a State Corporation Commission order, sought by the administration, requiring data centers to pay the full cost of transmission infrastructure built exclusively to serve them rather than spreading those costs across all ratepayers.[36] Virginia’s Chief Energy Officer, Josephus Allmond, summarized the outcome:
“Now data centers will pay for those costs directly.” — Josephus Allmond, Chief Energy Officer, Commonwealth of Virginia [36]
At the same time, Virginia has maintained mechanisms supporting advanced-nuclear research, workforce development, and future SMR site selection through its Power Innovation Program, and Virginia Energy has solicited industry information on potential nuclear projects and commercial offtakers. The lesson is significant, and it generalizes. Virginia’s future nuclear strategy does not have to mean placing a reactor next to every Northern Virginia data center. It can mean developing firm generation elsewhere in the Commonwealth — Southside, Southwest, former coal and industrial regions seeking replacement economic anchors — strengthening the transmission that links them, and using the state’s existing nuclear workforce and institutions (Dominion’s North Anna and Surry stations, the naval-nuclear complex in Hampton Roads, university programs) to support rising demand. The AI economy thus has the potential to connect two very different Virginias: the digital concentration of the north and the energy-development ambitions of everywhere else.
2.6 Pennsylvania: AI Demand Meets Legacy Nuclear Geography — and the Grid Becomes the Critical Path
Pennsylvania demonstrates how existing nuclear geography can become newly valuable, and it supplies this paper’s most important cautionary tale.
The Shapiro administration’s GRID Standards — the Governor’s Responsible Infrastructure Development framework, released in full in May 2026 and passed by the Pennsylvania House in June — condition state benefits for large data-center projects on enforceable commitments: developers must present an energy plan showing how they will meet their demand without imposing costs on other ratepayers, must build, bring online, or buy the incremental capacity their load requires while paying its full cost, must source rising percentages of that capacity from in-state dispatchable clean firm resources (up to 32 percent by 2035), and must pay all interconnection, transmission, and network-upgrade costs their demand causes.[37][38] It is among the most explicit statements yet by any state that AI growth and ratepayer protection will be linked by law rather than by hope — and its clean-firm escalator is, in effect, a standing procurement signal for precisely the kind of generation nuclear provides.
The state has simultaneously celebrated the marquee assets of its AI-energy economy: Amazon’s multibillion-dollar data-center investments and, above all, the restart of the Crane Clean Energy Center — the former Three Mile Island Unit 1 — underwritten by Microsoft’s twenty-year power purchase agreement and a $1 billion DOE loan, with more than 400 employees hired and a restart targeted for 2027, a year ahead of the original schedule.[13]
Yet Crane also delivers the caution. In early 2026, grid operator PJM indicated that the transmission upgrades needed for full deliverability of the plant’s 835 megawatts — including 765-kV and 500-kV projects — were not expected in service until December 2030 or later, implying that a reactor ready in 2027 might not achieve full grid integration until approximately 2031.[13][14] Constellation sought waivers to transfer capacity interconnection rights from a retiring fossil plant, drawing opposition from PJM’s independent market monitor before securing federal relief, and CEO Joseph Dominguez pushed back publicly on the most pessimistic readings:
“I don’t want anybody to be under the misconception the plant won’t start sooner.” — Joseph Dominguez, CEO, Constellation Energy [15]
The plant, in other words, will generate before the grid can fully absorb it. This is perhaps the clearest example in the country of why Reactor Readiness cannot mean reactor readiness alone. A reactor can be technically ready; the grid may not be. The offtaker may be ready; the transmission may not be. Capital may be available; the interconnection may not be. Real Reactor Readiness requires the complete system to mature together — and Section 3 builds that requirement into the framework itself.
Section 3: A Reactor Readiness Framework for States and Communities
If states are going to compete on readiness, they need a way to think about it systematically — a shared vocabulary that lets a governor’s office, a county commission, a utility, a developer, and an investor evaluate the same place and reach compatible conclusions. This section proposes a five-part Reactor Readiness Framework, and then converts it into a 100-point Reactor Readiness Index suitable for comparing states, counties, or specific sites.
The five dimensions are: (1) Site and Infrastructure Readiness; (2) Grid and Power-System Readiness; (3) Regulatory and Institutional Readiness; (4) Workforce and Supply-Chain Readiness; and (5) Community and Commercial Readiness. No single dimension is sufficient. The strength of the framework comes from their interaction — and, as the Crane case demonstrated, its failures come from allowing one dimension to lag years behind the others.
3.1 Site and Infrastructure Readiness
Before any discussion of reactor technology, communities need credible sites, and credibility is established by answers to a long list of unglamorous questions. Is the land controlled — owned, optioned, or assembled — rather than merely identified on a map? What is the geology, and what seismic considerations apply? What water resources are available for cooling, and under what riparian or groundwater regime? Can heavy components — reactor vessels weighing hundreds of tons — actually reach the site by road, rail, or water? How close is high-voltage transmission, and at what voltage class? Is there adequate laydown and construction space? What environmental sensitivities, wetlands, or protected habitats exist? Are there brownfield industrial facilities nearby whose infrastructure can be reused? What emergency-planning infrastructure — sirens, evacuation routes, hospital capacity, trained responders — already exists? And, least discussed but often decisive: does the community possess enough housing, schools, and public services to absorb a temporary construction workforce that may number in the thousands?
These questions are not glamorous, but they determine whether a theoretically attractive project is executable. Two categories of sites deserve systematic attention. Retired fossil-generation sites are particularly valuable because many already possess transmission interconnections, industrial zoning, water access, rail connections, skilled workers, and communities culturally familiar with energy production — the logic underlying TerraPower’s coal-transition siting in Wyoming and a substantial academic literature on coal-to-nuclear conversion. Existing nuclear sites are more valuable still, because they add licensing history, security perimeters, emergency-planning zones, and operational workforces to the bundle. Palisades demonstrates the enormous advantage of preserving nuclear campuses rather than allowing them to become economically inert; Holtec’s ability to propose two SMR-300s at the same site rests entirely on assets that would have been bulldozed under the original decommissioning plan.[16][19]
3.2 Grid and Power-System Readiness
This may be the most underestimated dimension of the five, and the events of 2026 in Pennsylvania have made it impossible to ignore. An advanced reactor does not create economic value merely by producing electricity; the electricity must reach customers, and the pathway runs through interconnection rights, transmission capacity, substations, transformers, protection systems, market rules, utility planning processes, and often years of regional transmission-organization study cycles.
The Crane situation shows how transmission can become the critical path even when the generating facility itself is advancing ahead of schedule: an 835-megawatt unit targeted for 2027 operation confronting network upgrades with in-service dates of December 2030 and beyond, requiring waiver litigation at FERC merely to deliver its full output.[13][14] AI changes this calculation further because some future customers will consume power at unprecedented scale and concentration; PJM’s own capacity auctions have already transmitted data-center scarcity into consumer prices across thirteen states, making transmission adequacy a political question as much as an engineering one.
A state considering advanced nuclear should therefore map not only where reactors could be built, but where electricity can realistically be delivered to projected AI and industrial loads. This suggests a new planning unit for state energy offices: the reactor + transmission + load corridor. The plant cannot be evaluated independently of the infrastructure connecting it to the economic activity it is intended to support. States that align generation siting, transmission planning, and large-load siting into common corridors — and that push their regional transmission organizations to study those corridors proactively — will compress timelines in ways no subsidy can replicate. Co-location and behind-the-meter configurations, flexible interconnection of the kind the Princeton ZERO Lab work describes, and phased energization agreements all belong in this dimension’s toolkit.[40]
3.3 Regulatory and Institutional Readiness
Nuclear development crosses an unusual number of institutional boundaries. The NRC controls core federal nuclear-safety licensing, but states and localities influence land use, water permits, environmental reviews, road use and heavy-haul permits, local taxation, construction approvals, economic incentives, public-utility regulation, workforce programs, and emergency management. Fragmentation across these authorities creates delay that has nothing to do with safety — a developer’s year spent discovering which agency controls which approval protects no one.
Texas is attempting to solve exactly this problem through TANEO, which serves as a single point of contact and liaison, guiding developers through state-level permitting and coordinating across agencies without displacing any agency’s substantive authority.[27] The model deserves wide replication, with one crucial design principle: the objective is not less review; it is less administrative uncertainty. Those are different things. Predictability reduces project cost and financing risk without compromising safety — indeed, a coordinated state process is more likely to surface genuine issues early, when they can be addressed cheaply, than a fragmented one in which issues emerge serially over years. Institutional readiness also includes quieter assets: state agencies with NRC-experienced staff, emergency-management programs familiar with radiological planning, a public-utility commission that has thought through cost-recovery and PPA structures for advanced nuclear in advance of the first application, and standing coordination channels with DOE programs — from the Gen III+ pathway to the supply-chain loan facilities — whose money increasingly flows to prepared applicants.[23][25]
3.4 Workforce and Supply-Chain Readiness
Reactors cannot be mass-deployed if every project must reinvent its workforce. Advanced nuclear requires expertise across nuclear engineering, welding to nuclear codes, quality assurance under NQA-1, nuclear-grade concrete, electrical construction, instrumentation and controls, reactor operations, cybersecurity, radiation protection, fuel handling, advanced manufacturing, project management, and regulatory compliance — and the national buildout will be limited less by reactor designs than by whether the United States can manufacture enough qualified components and train enough people to build multiple reactors simultaneously.
States should therefore begin workforce development before final investment decisions, not after. Community colleges can create nuclear-construction and technician programs; universities can expand reactor engineering; building-trades unions can establish specialized training pipelines; existing coal-plant and nuclear workers can be retrained; regional manufacturers can obtain nuclear quality certifications; and local suppliers can learn the documentation and quality-control regimes of nuclear procurement, which differ from ordinary industrial work more in paperwork discipline than in technical difficulty. This is the dimension where Tennessee’s ecosystem is most instructive — the state is deliberately manufacturing the inputs to everyone else’s readiness — and where the federal supply-chain actions are most complementary: DOE’s $17.5 billion long-lead financing and its Gen III+ awards to forgers, fabricators, and fuel lines explicitly target the same constraint, with components expected to be sourced from over 100 domestic companies across 40 states.[25][26][23][32]
3.5 Community and Commercial Readiness
This final dimension combines two factors usually discussed separately but that increasingly belong together: a reactor needs a host community, and it needs a customer, and the two relationships shape one another.
Communities will reasonably ask: Who benefits? Who pays? How many permanent jobs will exist? Will residential electricity rates rise? What happens to water resources? Who finances roads and transmission? What emergency responsibilities fall on local government? How will tax revenues be shared? What happens if the data-center project is canceled? What happens if the reactor is delayed? These questions should not be treated as obstacles appearing at the end of project development; they should be incorporated at the beginning, because a community that knows what it expects can negotiate quickly and credibly, while a community encountering these questions for the first time under deadline pressure will either refuse or extract terms that damage the project. The strongest readiness strategies therefore establish community-benefit principles before developers arrive — Indiana’s tiered program effectively institutionalizes exactly this sequencing.[1][2]
On the commercial side, reactor developers increasingly need credible long-term electricity buyers, and this is where hyperscalers enter the framework. AI companies possess enormous electricity demand and unusually strong balance sheets; their long-term contracts can give developers the revenue certainty that merchant electricity markets rarely provide, as Section 4 details. Community readiness and commercial readiness thus meet around the same project: the community-benefit agreement and the power purchase agreement are two halves of the same social-and-commercial license.
3.6 The Reactor Readiness Index
To make the framework operational, I propose a 100-point Reactor Readiness Index, allocated to reflect both the relative cost of remedying deficits and the degree to which each dimension gates the others.
| Dimension | Points | Illustrative Indicators |
| 1. Site & Infrastructure | 25 | Land control; geology/seismic screening; water rights; heavy-haul access (road/rail/barge); laydown area; brownfield/retired-plant/existing-nuclear reuse; emergency-planning assets; construction-workforce housing |
| 2. Grid & Power System | 20 | Distance to high-voltage transmission; interconnection position and CIR availability; substation/transformer headroom; RTO study status; identified reactor–transmission–load corridor; co-location/flexible-interconnection options |
| 3. Regulation & Institutions | 20 | Single-point state coordination office; mapped state/local permitting sequence; NRC-experienced state staff; emergency-management readiness; PUC frameworks for advanced-nuclear cost recovery; active DOE program engagement |
| 4. Workforce & Supply Chain | 20 | Nuclear-qualified trades pipeline; community-college and university programs; NQA-1 certified in-state suppliers; fuel-cycle access; proximity to nuclear manufacturing clusters; retainable coal/nuclear workforce |
| 5. Community & Commercial | 15 | Public education record; local resolutions and zoning posture; pre-negotiated community-benefit principles; identified prospective offtakers; tax and cost-allocation structures; political durability across election cycles |
| Total | 100 |
Table 3. The Reactor Readiness Index: a 100-point scoring architecture.
Three features of the design deserve comment. First, the weights are deliberately front-loaded toward physical and grid factors (45 of 100 points) because these are the slowest and most expensive to remedy: a missing transmission corridor takes the better part of a decade to build, while a coordination office can be established in a legislative session. Second, the index is intended to be scored at the site or county level and aggregated upward, because state averages conceal exactly the local variation that matters — Tennessee would score modestly as an “average” but extraordinarily at Oak Ridge. Third, the index measures readiness, not announcements: a signed memorandum of understanding scores nothing; a controlled site with a completed transmission study scores a great deal. Section 5 returns to this distinction as a management principle. The immediate uses are practical: states can self-assess to direct scarce program dollars at their lowest-scoring gating dimension; developers can screen jurisdictions consistently; and federal programs could, in time, condition cost-shared awards partly on demonstrated readiness, multiplying the leverage of every federal dollar.
Section 4: Hyperscalers Are Becoming Nuclear Market Makers
4.1 The Buyer Is Changing
For most of nuclear history, regulated utilities decided whether reactors would be built, because only utilities combined the demand visibility, the balance sheet, and the regulatory cost-recovery mechanisms the technology’s capital intensity required. That model is changing before our eyes. Technology companies are becoming principal participants in nuclear economics — not as operators, but as the providers of the single input reactor developers most desperately need: credible long-term demand.
Meta’s January 2026 announcements illustrate the scale of the transition. Its agreements with Vistra, TerraPower, and Oklo — layered on the 2025 Constellation contract for the Clinton plant — are designed to preserve existing nuclear production, fund plant uprates, and underwrite entirely new advanced-reactor construction, supporting up to 6.6 gigawatts by 2035, all within PJM, and framed by the company as making it one of the most significant corporate purchasers of nuclear energy in American history.[10] The structure is worth unpacking, because each leg performs a different market function. The Vistra leg comprises twenty-year power purchase agreements for more than 2,100 megawatts from the Perry, Davis-Besse, and Beaver Valley stations, plus backing for 433 megawatts of uprates — life-extension and expansion of existing assets.[11] The Oklo leg funds fuel procurement and first-phase development of a 1.2-gigawatt campus of Aurora powerhouses in Pike County, Ohio, targeted to begin producing as early as 2030 — new advanced capacity called into existence by the contract.[11] The TerraPower leg funds development of two 345-megawatt Natrium units for delivery after 2032 — a multi-unit order for a design whose first unit is still under construction in Wyoming.[11] TerraPower’s chief executive, Chris Levesque, stated the industrial logic directly:
“We must deploy gigawatts of advanced nuclear energy in the 2030s.” — Chris Levesque, President and CEO, TerraPower [51]
A technology company that behaves this way is no longer merely an electricity customer. It is a market creator for new generation — performing, through bilateral contract, the demand-aggregation function that rate-based utility planning once performed through regulation.
4.2 The Nuclear Offtake Contract as Infrastructure Finance
An advanced reactor is extraordinarily capital-intensive, and investors therefore care above all about future revenue certainty. A twenty-year electricity agreement from an investment-grade hyperscaler transforms project economics: it supports debt, anchors supply-chain orders, justifies workforce commitments, and gives every other participant in the capital stack confidence that the electricity will have a buyer at a known price. Meta itself described the mechanism candidly — the agreements give Oklo and TerraPower “greater business certainty” and the ability to “raise capital to move forward.”[10] The federal government has now built the same logic into public finance: DOE officials structuring the $17.5 billion supply-chain loans indicated the supported AP1000 projects would rest on hyperscaler power purchase agreements of up to twenty-five years, priced to ensure projects proceed without burdening local ratepayers.[26]
This introduces a striking historical reversal. For a century, electricity infrastructure enabled the emergence of new industries; the grid came first, and the aluminum smelter, the assembly line, or the server farm followed. Increasingly, the causation also runs the other way: commitments from the digital economy are enabling electricity infrastructure that would not otherwise be financed. The relationship becomes circular — AI demand → nuclear contract → reactor financing → additional firm electricity → larger AI infrastructure — and, managed well, virtuous. None of this eliminates construction risk; first-of-a-kind projects will still overrun and slip. But it redistributes risk across reactor developers, utilities, states, federal credit programs, infrastructure investors, and technology companies with the balance sheets to absorb it — which is precisely what the sector’s traditional structure could no longer do.
4.3 Amazon, X-energy, and the Industrialization of Advanced Nuclear
Amazon’s backing of X-energy demonstrates a complementary pathway: equity investment upstream of any specific power plant, aimed at industrializing a technology platform. X-energy — developer of the Xe-100 high-temperature gas-cooled reactor and the TRISO-X fuel line — completed its initial public offering in April 2026, raising over $1 billion after Amazon’s earlier investment rounds, and reported its first full public quarter in August 2026 with revenue and grant income up 154 percent, $1.9 billion of cash, binding HALEU enrichment agreements with Centrus and General Atomics, and an investment to double SGL Carbon’s nuclear graphite capacity by 2030.[49][5]
Critically, the company’s first deployment is industrial rather than digital: four Xe-100 units supplying power and high-temperature steam to Dow’s Seadrift petrochemical complex in Texas, now supported by up to $2.15 billion in federal cost share and targeted for the early 2030s as North America’s first grid-scale advanced reactor serving an industrial site.[4] That distinction is analytically useful. The advanced-nuclear supply chain does not care whether its learning curve is financed initially by a chemical plant, a military installation, a utility, a hyperscaler, or a semiconductor fab. Every completed project reduces cost and schedule uncertainty for all subsequent ones. AI will therefore benefit from reactors that were initially justified by entirely different industries — one more reason state Reactor Readiness programs should never be designed exclusively around technology companies.
4.4 Bill Gates, TerraPower, and the Internationalization of the Capital Stack
TerraPower supplies a third pattern: founder capital meeting sovereign industrial strategy. Bill Gates co-founded the company in 2008, long before the AI electricity boom made advanced nuclear fashionable, and its Natrium project in Kemmerer, Wyoming — a 345-megawatt sodium-cooled fast reactor paired with molten-salt thermal storage that can flex output to 500 megawatts — received its NRC construction permit in March 2026, the first in decades for a commercial non-water-cooled reactor, with commercial operation targeted for 2031.[6][7]
The August 14, 2026 events in Seoul then revealed how international the capital stack has become. SK Innovation — already, with SK Inc., TerraPower’s second-largest shareholder following a $250 million investment in 2022 — signed the term sheet for a global joint business covering Korea’s first commercial Natrium plant and joint pursuit of projects worldwide, with completion, price, and performance guarantees intended to unlock conventional commercial financing for a fleet.[6][7] Doosan Enerbility took the manufacturing contract for the Natrium reactor vessel and internals; HD Hyundai deepened its own sodium-fast-reactor memorandum; and Gates spent the day shuttling among chaebol chairmen and ministers discussing power for AI.[48] The significance extends well beyond one company: nuclear development is becoming international industrial policy. An American reactor design may rely on Korean heavy manufacturing, Canadian uranium, American enrichment, specialized European components, federal credit support, state incentives, and hyperscaler offtake — simultaneously. Reactor Readiness at the state level therefore eventually connects to supply-chain diplomacy at the national level, and states with ports, heavy-haul corridors, and foreign-direct-investment programs hold cards they may not realize they hold.
4.5 The Hyperscaler Nuclear Portfolio, Mid-2026
| Company | Counterparties | Structure | Scale & Horizon |
| Microsoft | Constellation (Crane Clean Energy Center, PA) | 20-year PPA underwriting first U.S. restart of a retired large reactor | 835 MW; restart targeted 2027; full grid deliverability contested to ~2031 [13] |
| Meta | Constellation (Clinton); Vistra (Perry, Davis-Besse, Beaver Valley); Oklo (Pike County, OH); TerraPower | 20-year PPAs; uprate backing; development funding for new advanced units | Up to 6.6 GW by 2035, all in PJM [10][11] |
| Amazon | X-energy (equity; IPO anchor); utility partnerships | Platform equity + project development; TRISO fuel chain | Xe-100 fleet ambitions; first units at Dow Seadrift, early 2030s [4][49] |
| Kairos Power | Multi-reactor development agreement for molten-salt SMRs | Fleet-scale intent, 2030s | |
| Federal AP1000 program (hyperscaler-backed) | Westinghouse + utilities, DOE EDF | $17.5B conditional long-lead loans; PPAs up to 25 years anticipated with hyperscalers | 10 large reactors, 5 sites; construction goal by 2030 [25][26] |
Table 4. Hyperscaler and hyperscaler-adjacent nuclear commitments through August 2026.
4.6 The AI Company May Arrive After the Reactor Decision
This leads back to the title of this paper. The traditional assumption ran: data center arrives → electricity demand appears → reactor is proposed. A Reactor Readiness strategy contemplates the opposite sequence: community prepares → state organizes → site qualifies → reactor developer engages → generation project advances → hyperscaler chooses the location.
That reversal is potentially transformational for state economic development. If a state can demonstrate that several hundred megawatts — or several gigawatts — of firm generation can realistically become available on a credible schedule, electricity supply itself becomes a site-selection incentive, arguably the most valuable one now on offer. Instead of asking, how will we power the data center that wants to come here?, a governor may eventually ask, which AI company wants to locate next to the power capacity we have already prepared? That is a fundamentally stronger bargaining position — stronger on price, stronger on community benefits, stronger on cost allocation — and it is available only to the prepared.
Section 5: Building a Reactor-Ready State — A Policy Roadmap for 2026–2035
The framework of Section 3 tells a state what readiness consists of. This section tells it what to do, in sequence, over the decade in which the American nuclear-AI buildout will largely be decided. Ten steps follow, summarized in Table 5 and elaborated below. They are ordered deliberately: inventory before institutions, institutions before incentives, cost discipline before recruitment.
| Step | Action | Primary Actor | Horizon |
| 1 | Inventory the nuclear geography | State energy office + GIS/utility partners | 2026–2027 |
| 2 | Build a state nuclear coordination office | Governor + legislature | 2026–2027 |
| 3 | Create community readiness programs | State energy office + localities | 2026–2028 |
| 4 | Couple reactor policy with large-load policy | Legislature + PUC | 2026–2028 |
| 5 | Protect ratepayers with explicit cost allocation | PUC + legislature | 2026–2028 |
| 6 | Prepare the workforce years before operation | Education systems + trades + utilities | 2026–2032 |
| 7 | Build the nuclear supply chain as a cluster | Economic development + manufacturers | 2026–2033 |
| 8 | Connect readiness to national security | State + federal installations + DOE | 2027–2035 |
| 9 | Measure readiness, not announcements | Governor’s office + independent scorer | Continuous |
| 10 | Treat time as an economic asset | All of the above | Continuous |
Table 5. The ten-step Reactor Readiness roadmap, 2026–2035.
5.1 Step One: Inventory the Nuclear Geography Before Offering Incentives
Every state considering advanced nuclear should begin with a physical inventory, because incentives offered before geography is understood are incentives mispriced. The inventory should identify existing nuclear sites; retired and retiring fossil plants; large substations and major transmission corridors; industrial brownfields; federal facilities and military bases; large water sources; rail and heavy-haul connections; energy-intensive industrial clusters; uranium and fuel-cycle facilities; and regions actively seeking replacement economic anchors. It should then be overlaid with future-load forecasts — utility integrated resource plans, RTO queue data, announced and rumored data-center projects — because the objective is not simply to locate possible reactor sites. It is to identify where future generation and future demand can realistically meet: the reactor–transmission–load corridors of Section 3.2. States that publish a screened, honest version of this inventory will find that it functions as a developer-attraction document more powerful than any brochure.
5.2 Step Two: Build a State Nuclear Coordination Office
Texas’s TANEO offers the template: a nonregulatory office providing strategic leadership, a single point of contact, and agency-by-agency navigation for developers, paired with an administering role over state funding.[27] A well-built office should house expertise spanning energy, economic development, environmental regulation, emergency management, workforce, transportation, public utilities, local government, higher education, and federal nuclear policy — because those are the boundaries a project actually crosses. Its performance metric should not be the number of press releases issued; it should be measurable reductions in unnecessary project uncertainty: days from inquiry to complete permitting map, number of pre-application coordination meetings convened, cycle time on state-controlled approvals. Smaller states need not replicate Texas’s scale; even a three-person office with genuine convening authority changes developer behavior, because it converts an opaque institutional landscape into an addressable one.
5.3 Step Three: Create Community Readiness Programs Before Site Selection Becomes Political
Indiana’s innovation is the most replicable single element in this paper. Instead of waiting for a specific nuclear proposal to become a local controversy — the moment when positions harden and information arrives pre-polarized — a state can invite communities to educate themselves voluntarily, on their own schedule, with no project on the table.[1][2] The process can include public nuclear-energy education; preliminary site screening; local resolutions; emergency-services assessment; workforce surveys; tax and fiscal modeling; water analysis; infrastructure planning; developer workshops; and early community-benefit discussions. Two design principles are essential. First, participation must never guarantee that a reactor will be approved; it means only that the community understands what hosting one would involve. That distinction preserves democratic legitimacy — readiness is the capacity to say yes quickly, which includes the capacity to say no early and cheaply. Second, the credential must be meaningful: tiers should require verifiable work products (a completed site screen, an executed education program, a workforce partnership agreement), because a designation that can be obtained by resolution alone will be discounted by exactly the developers it is meant to attract.
5.4 Step Four: Couple Reactor Policy with Large-Load Policy
States should refuse to run nuclear programs and data-center programs in separate policy silos, because the two are now the supply and demand sides of a single market. Michigan’s 2026 approach — supporting the Palisades restart and future SMRs while moving to codify safeguards that make data centers responsible for their own costs — is directionally exemplary.[30][31] Pennsylvania’s GRID Standards go further, converting data-center benefit eligibility into a standing procurement mechanism for in-state dispatchable clean firm power, escalating to 32 percent by 2035 — a demand-pull instrument for exactly the generation class nuclear occupies.[37] Virginia’s consumption tax and SCC transmission-cost order complete the pattern from the cost side.[34][35][36] Together these policies suggest an emerging interstate bargain: states will help create the generation and infrastructure the AI economy requires, and AI developers will increasingly be expected to bear the costs they impose on the system. Nuclear fits naturally within that bargain, because its output profile matches what the standards demand and its financing benefits from precisely the long-term corporate commitments the bargain elicits.
5.5 Step Five: Protect Ratepayers — the Decisive Political Requirement
This may become the make-or-break condition for the entire agenda. AI infrastructure is economically attractive: investment, tax revenue, construction activity, technology employment, strategic importance. But voters experience the electricity system through their monthly bill, and 2026 has already demonstrated — in PJM capacity prices, in state legislative sessions, in gubernatorial messaging from Richmond to Lansing to Harrisburg — that AI-driven cost shifts are becoming a first-order political issue. Reactor Readiness should therefore include explicit, legislated answers to the allocation questions before the first project arrives: who pays for transmission; who pays for generation; who assumes construction risk; who bears cancellation risk; who pays for local infrastructure; and what happens if the anticipated data-center load never materializes. A nuclear project justified by a private AI buyer should not automatically transfer its development risks onto captive utility customers. The Virginia SCC’s August 2026 order — data centers pay the full cost of transmission built exclusively for them — and the equity-first structure of DOE’s supply-chain loans both point the same direction.[36][26] Strong cost discipline is not an obstacle to nuclear-AI development; it is what makes such development politically durable across election cycles, which is itself a readiness asset under Dimension 5 of the Index.
5.6 Step Six: Prepare the Workforce Years Before Commercial Operation
A state expecting advanced reactors in the early 2030s should be training workers now, because the training pipelines are longer than the remaining runway. Programs should connect high schools, community colleges, engineering universities, building trades, utilities, reactor developers, national laboratories, manufacturers, and military veterans — the last a particularly rich source of nuclear-trained talent. A young person beginning technical education in 2026 can stand inside the containment of a first-wave commercial advanced reactor as a journeyman in 2032. States that instead wait for construction contracts before training will import labor at premium cost, forfeit the local economic benefits that justified community support in the first place, and — because craft labor availability is now a national constraint — may watch their projects slip behind states that trained early. Workforce readiness is also the dimension most visible to communities: a nuclear-technology program at the county community college does more for local acceptance than any advertising campaign, because it makes the future tangible and locally owned.
5.7 Step Seven: Build the Nuclear Supply Chain as an Economic Cluster
The largest prize may not be the reactor itself. A state that develops exportable expertise in reactor vessels, heat exchangers, specialized pumps, TRISO fuel, HALEU enrichment, instrumentation and control systems, nuclear-grade concrete, robotic inspection, digital twins, cybersecurity, remote operations, or advanced manufacturing can capture economic value from projects across the entire national — and international — buildout. Tennessee is executing this strategy explicitly, and the returns compound: each Oak Ridge announcement makes the next more likely, because suppliers locate near suppliers.[32][33] Texas has written supply-chain development directly into its fund’s eligible uses.[27] The federal complement is enormous — $17.5 billion in long-lead financing flowing to components sourced from over 100 companies in 40 states means nearly every state already hosts a latent nuclear supplier that a modest certification grant could activate.[25][121] The advanced-nuclear economy could eventually resemble semiconductor manufacturing, where the visible finished product depends on a deep ecosystem of specialized suppliers, and where the durable rents accrue disproportionately upstream.
5.8 Step Eight: Connect Reactor Readiness to National Security
Nuclear-powered AI infrastructure is also becoming a national-security matter, and states should read the federal signals accordingly. The May 2025 executive orders on advanced reactors explicitly connected nuclear power to the electricity requirements of AI computing and critical defense infrastructure, and set the criticality and construction milestones that the Reactor Pilot Program and supply-chain loans now execute.[20][46] The June 2026 National Security Presidential Memorandum on artificial intelligence went further, directing the development within ninety days of a roadmap ensuring the national-security enterprise has adequate access to advanced computing — including the commissioning of high-security AI computing facilities operating at scale.[45] Facilities of that kind will require firm, secure, and preferably on-site or dedicated power. This creates a foreseeable category of projects where AI compute, advanced nuclear, military installations, national laboratories, secure cloud infrastructure, and federal energy policy converge. States containing major defense installations or national laboratories — Idaho, Tennessee, New Mexico, Virginia, Texas, South Carolina, Washington — should recognize the pattern early and prepare the surrounding readiness assets, because federal siting decisions, once made, anchor decades of regional investment.
5.9 Step Nine: Measure Readiness, Not Announcements
States are fluent in celebrating announced investments; readiness requires more austere accounting. The Reactor Readiness Index of Section 3.6 supplies the instrument: score sites and counties annually across the five dimensions, publish the methodology, and let the scores — not the press releases — direct program dollars toward the lowest-scoring gating dimension. The discipline matters because announcements and readiness can diverge for years: a state can accumulate memoranda of understanding while its best site remains without land control, or celebrate a designation program while its transmission corridors go unstudied. An honest index makes such divergences visible while they are still cheap to fix. It also enables the comparison that motivates the whole exercise: distinguishing a state that merely says it supports nuclear from one that could actually host a project on a customer’s schedule — a distinction that developers, investors, and hyperscaler site-selection teams are already drawing privately, whether or not states measure it publicly.
5.10 Step Ten: Treat Time as an Economic Asset
The final recommendation is the most important, because it is the economic logic underneath the other nine. AI hardware depreciates rapidly; the accelerators inside a gigawatt campus lose competitive edge within a handful of years and are refreshed on cycles far shorter than any power plant’s construction schedule. A delayed reactor therefore imposes an opportunity cost on an AI customer that is wholly out of proportion to the electricity’s price: the value of the computing that could not run. Against capital commitments the size of 2026’s hyperscaler budgets — roughly $725 billion across four companies in a single year — reducing power-availability uncertainty by even one or two years is worth billions of dollars per campus.[43][44] That means Reactor Readiness creates an economic asset that appears on no state balance sheet: time saved. States that internalize this will eventually compete not primarily on the size of their subsidy packages but on the credibility of their timelines — and credibility, unlike subsidy, compounds. Every project a prepared state delivers on schedule lowers the risk premium on the next one.
Section 6: What Have We Learned? Seven Pillars
The argument of this paper can be compressed into seven pillars — five drawn from the structural analysis, and two that emerged from the events of 2026 themselves.
Pillar 1 — Prepare the Energy Ecosystem Before the AI Load Arrives
The first lesson is that energy policy can no longer remain reactive. For decades, utilities could respond to new commercial demand after customers chose where to locate, because loads were small and grids had headroom. Gigawatt-scale AI infrastructure ends that era. If states wait until a giant campus is announced before asking how to supply the electricity, they will discover that transmission, permitting, generation, and workforce timelines are incompatible with the customer’s schedule — and the customer will not wait. Reactor Readiness reverses the sequence: prepare first, recruit second. A prepared energy ecosystem becomes part of the reason the AI investment arrives at all.
Pillar 2 — A Reactor Is Only as Ready as the System Around It
The second lesson is that reactor technology by itself is insufficient, and 2026 furnished the proof in every direction. Crane demonstrates the primacy of transmission — a plant ready in 2027 confronting network upgrades dated 2030 and beyond.[13][14] Palisades demonstrates the depth of operational and regulatory restart readiness — a “watershed” milestone still trailed by five thousand work activities.[16][17] Indiana demonstrates the necessity of communities.[1] Texas demonstrates capital, coordination, and supply chains.[27] Tennessee demonstrates manufacturing and fuel.[32] The unit of analysis must therefore expand beyond the reactor to the full system: Reactor + Grid + Site + Workforce + Community + Customer + Capital. If any single element falls several years behind, the entire project inherits the delay.
Pillar 3 — Communities Are Becoming Strategic Infrastructure
The third lesson is the easiest to overlook because it involves no hardware. A community that understands nuclear development is itself an infrastructure asset. Political trust cannot be ordered from a factory; public legitimacy cannot be delivered on a truck; local knowledge cannot be manufactured overnight. If advanced nuclear deployment expands substantially in the 2030s, the United States may discover that one of its scarcest resources is not uranium or reactor steel but communities willing and institutionally prepared to host projects. Indiana’s program deserves attention far beyond Indiana precisely because it treats community preparedness as something that can be deliberately, measurably, and inexpensively developed in advance.[1][2]
Pillar 4 — Hyperscalers Are Becoming Part of the Nuclear Capital Stack
The fourth lesson is that technology companies are no longer passive electricity consumers. Microsoft’s contract resurrects a retired reactor; Meta’s portfolio spans preservation, uprates, and new construction across three counterparties; Amazon industrializes a reactor developer through equity and an IPO; Google underwrites a fleet pathway; and the federal government now structures $17.5 billion of public credit on the assumption of hyperscaler offtake.[13][10][49][25][26] The hyperscaler has become offtaker, investor, project catalyst, creditworthy counterparty, political stakeholder, and — in aggregate — a driver of which generation assets get financed at all. This links Layer 1 of the Five-Layer AI Economy directly to Layers 3, 4, and 5: the AI company’s future demand now helps determine the present capital allocation of the power sector.
Pillar 5 — Reactor Readiness Is Becoming a Form of State Competitiveness
The fifth lesson brings the structure together. States once competed for technology investment through taxes, land, universities, fiber, transportation, and labor. They now increasingly compete through electricity, and the next stage is competition through future electricity certainty. A state that can say only we have land, but no power plan is weaker than a state that can say: we have identified the site, the transmission corridor, the reactor options, the workforce, the community partners, the regulatory pathway, the supply-chain partners, and the cost-allocation structure required to deliver firm power on a schedule you can finance against. That difference is Reactor Readiness, and it may decide where America’s next generation of AI factories is built.
Pillar 6 — The Ratepayer Bargain Is the License to Operate
The sixth pillar emerged unmistakably from the 2026 state legislative and regulatory record. Virginia taxed data-center consumption and shifted dedicated transmission costs onto the facilities that cause them; Michigan demanded that working families “not pay a single penny” for data-center development; Pennsylvania conditioned benefits on developers paying the full cost of the capacity and upgrades their load requires.[34][36][31][37] The pattern is bipartisan in effect if not always in rhetoric, and it defines the political envelope within which the nuclear-AI buildout will proceed: large loads pay their way, or the buildout loses its public license. Far from threatening nuclear development, this bargain advantages it — because nuclear’s natural financing structure, the long-term corporate PPA, is precisely the instrument that satisfies the bargain without touching the residential bill.
Pillar 7 — Federal Acceleration Rewards the Prepared
The seventh pillar is the compounding interaction between Washington and the states. The federal government spent the twelve months through August 2026 systematically shortening the reactor side of the timeline: four criticalities by Independence Day, $900 million for Gen III+ deployment, $17.5 billion for long-lead components, cost-share expansions for Generation IV demonstrations.[20][21][23][25][4] Every one of those instruments flows toward prepared applicants and prepared places — Tier 1 awards went to a state with an early site permit in hand and a company with a nuclear campus to reuse; supply-chain loans require utilities ready to commit half a billion dollars of equity per project; pilot-program reactors went critical fastest where labs, regulators, and sites were already aligned. Federal acceleration is therefore not a substitute for state readiness. It is a multiplier on it — and multiplication by zero remains zero.
Conclusion: Reactor Readiness Comes Before Reactor Deployment
The great AI infrastructure race is usually described through its visible objects: GPUs, server racks, data centers, transmission lines, substations, cooling towers, nuclear reactors. But the most important infrastructure is sometimes what exists before any of those objects are built. A prepared institution. A trained workforce. A qualified site. A mapped transmission connection. A supportive community. A predictable permitting process. A financing mechanism. A credible electricity buyer. A political agreement explaining who pays. Together, those conditions determine whether a reactor concept can become an operating asset — and whether it can become one on a schedule that the fastest-moving industry in economic history is willing to build against.
It is worth re-emphasizing, in closing, why Reactor Readiness is the appropriate title for this paper. The phrase does not describe a nuclear technology. It describes a state of preparation. That distinction matters because the central nuclear challenge of the AI era is not whether American engineers can design advanced reactors. The United States now has an expanding roster of companies pursuing large reactors, SMRs, microreactors, sodium-cooled and gas-cooled systems, and nuclear batteries; the four criticalities achieved under the Department of Energy’s Reactor Pilot Program by July 4, 2026, demonstrated that advanced-reactor experimentation has genuinely accelerated.[20][21][46][50] The harder question is whether the surrounding economy can repeatedly convert those technologies into commercial plants. Can communities host them? Can states permit around them efficiently? Can manufacturers build components at scale? Can fuel suppliers expand? Can utilities connect them? Can workers construct and operate them? Can investors finance them? Can hyperscalers provide sufficient long-term demand? Can politicians protect ordinary ratepayers while still attracting enormous private investment? Can federal and state governments coordinate rather than collide? Those questions define readiness.
The emerging evidence suggests that some states have already understood the shift. Indiana is preparing communities before any specific project arrives. Texas is constructing a dedicated advanced-nuclear institution and the nation’s largest state funding mechanism. Michigan is recovering a retired nuclear asset while opening the same site to a next generation of reactors. Tennessee is building the fuel-cycle and manufacturing base on which every other state’s projects will draw. Virginia is binding the world’s densest data-center economy to cost responsibility while keeping its nuclear-innovation pathways open. Pennsylvania is pursuing AI investment, nuclear revival, and enforceable standards for large digital infrastructure simultaneously. These approaches differ — that is precisely their value as a natural experiment — but they share a common intuition: the nuclear race can be influenced long before concrete is poured.
That insight becomes more important still when placed inside the Five-Layer AI Economy. The energy layer is no longer simply an upstream supplier to the intelligence economy; it is becoming a constraint on how quickly the intelligence economy can grow. Roughly three-quarters of a trillion dollars of GPUs, data centers, models, and agentic applications planned for 2026 alone ultimately depend on the physical capacity to transform fuel, sunlight, wind, water, geothermal heat, natural gas, or nuclear fission into electrons.[43] As AI demand expands, states that possess reliable electricity will gain leverage. States that possess pathways for creating additional reliable electricity will gain even more. And states that prepare those pathways before everyone else recognizes their scarcity may acquire the greatest advantage of all.
That is the deeper meaning of Reactor Readiness. It is not a prediction that every AI data center will someday have a small modular reactor beside it; that outcome is neither necessary nor likely. It is the recognition that the jurisdictions most capable of hosting the AI economy of the 2030s will increasingly be those that prepared for enormous electricity demand before the demand arrived at their doorstep. The reactor may be built years later. The data center may belong to Meta, Amazon, Google, Microsoft, OpenAI’s infrastructure partners, xAI, or a company that does not yet exist. The reactor design may come from TerraPower, X-energy, Kairos, Oklo, Holtec, Westinghouse, GE Vernova Hitachi, or a startup whose test article went critical in a desert last June. Those choices can come later.
The preparation cannot.
That is Reactor Readiness: preparing the community, the grid, the workforce, the institutions, and the economic bargain before the first AI server — or the first reactor vessel — arrives.
Footnotes / Endnotes:
[1] Indiana Office of Energy Development, “Indiana Advanced Nuclear Ready Communities Program,” State of Indiana, 2026. https://www.in.gov/oed/indianas-energy-policy/nuclear-ready/
[2] American Public Power Association, “Indiana Governor Launches Advanced Nuclear Ready Community Program,” July 2026 (statements of Gov. Mike Braun and Sec. Suzanne Jaworowski). https://www.publicpower.org/periodical/article/indiana-governor-launches-advanced-nuclear-ready-community-program
[3] American Nuclear Society, Nuclear Newswire, “Indiana Launches Nuclear Readiness Designation System,” July 20, 2026. https://www.ans.org/news/2026-07-20/article-8216/indiana-launches-nuclear-readiness-designation-system/
[4] Timothy Gardner, Reuters (via EnergyNow), “X-Energy Gets Additional Up to $1 Billion Public US Funding for Texas Nuclear Project,” August 13, 2026. https://energynow.com/2026/08/x-energy-gets-additional-up-to-1-billion-public-us-funding-for-texas-nuclear-project/
[5] X-energy, Inc., Q2 FY2026 Earnings Call Transcript (Clay Sell, CEO), August 13, 2026. https://finance.yahoo.com/quote/XE/earnings/XE-Q2-2026-earnings_call-685748.html
[6] Reuters, “U.S. TerraPower, South Korea’s SK Innovation Sign Preliminary Deal on Global SMR Projects,” August 14, 2026. https://kfgo.com/2026/08/14/u-s-terrapower-south-koreas-sk-innovation-sign-preliminary-deal-on-global-smr-projects/
[7] World Nuclear News, “TerraPower Expands Cooperation with Korean Partners,” August 2026. https://www.world-nuclear-news.org/articles/terrapower-expands-cooperation-with-korean-partners
[8] U.S. Energy Information Administration, “Short-Term Energy Outlook,” August 2026. https://www.eia.gov/outlooks/steo/pdf/steo_full.pdf; see also Scott DiSavino, Reuters, “US Power Use to Beat Record Highs in 2026 and 2027 as AI Use Surges, EIA Says,” August 11, 2026. https://boereport.com/2026/08/11/us-power-use-to-beat-record-highs-in-2026-and-2027-as-ai-use-surges-eia-says-5/
[9] U.S. Energy Information Administration, “Data Center Server Energy Use Grows Across the Commercial Building Stock,” Today in Energy / Annual Energy Outlook 2026, May 19, 2026. https://www.eia.gov/todayinenergy/detail.php?id=67704
[10] Meta Platforms, Inc., “Meta Announces Nuclear Energy Projects, Unlocking Up to 6.6 GW to Power American Leadership in AI Innovation,” January 2026. https://about.fb.com/news/2026/01/meta-nuclear-energy-projects-power-american-ai-leadership/
[11] Ethan Howland, Utility Dive, “Meta Inks Nuclear Deals for Up to 6.6 GW from Oklo, Vistra, TerraPower,” January 9, 2026. https://www.utilitydive.com/news/meta-nuclear-deal-oklo-vistra-terrapower-ai-data-centers/809215/
[12] CNBC, “Meta Signs Nuclear Energy Deals to Power Prometheus AI Supercluster,” January 9, 2026. https://www.cnbc.com/2026/01/09/meta-signs-nuclear-energy-deals-to-power-prometheus-ai-supercluster.html
[13] World Nuclear News, “Constellation Seeks Regulator’s Help for 2027 Plant Restart,” April 2, 2026. https://www.world-nuclear-news.org/articles/constellation-seeks-regulators-help-for-2027-plant-restart
[14] Ethan Howland, Utility Dive, “PJM Market Monitor Opposes Waivers for Constellation’s Three Mile Island Nuclear Restart,” April 23, 2026. https://www.utilitydive.com/news/pjm-market-monitor-constellations-nuclear-crane-waiver/818216/
[15] American Nuclear Society, Nuclear Newswire, “FERC Decision on Crane Restart Coming in June or July, Constellation Execs Say” (Joseph Dominguez remarks), May 13, 2026. https://www.ans.org/news/2026-05-13/article-8026/ferc-decision-on-crane-restart-coming-in-june-or-july-constellation-execs-say/
[16] Holtec International, “Palisades’ Methodical March Toward Restart Reaches a Major Milestone,” July 2, 2026. https://holtecinternational.com/hh-41-10/
[17] American Nuclear Society, Nuclear Newswire, “Palisades: Restart Projects, Holtec IPO, Lawsuit Dismissal—But No Restart Date” (Kris Singh remarks to the Financial Times), July 8, 2026. https://www.ans.org/news/2026-07-08/article-8187/palisades-restart-projects-holtec-ipo-lawsuit-dismissal-but-no-restart-date/
[18] U.S. Nuclear Regulatory Commission, “Palisades Nuclear Plant” (restart oversight and 2026 site inspection activity plan). https://www.nrc.gov/info-finder/reactors/pali
[19] Dan Yurman, Neutron Bytes, “Holtec Submits License Application to NRC for the Palisades Twin SMR-300s,” January 11, 2026. https://neutronbytes.com/2026/01/11/holtec-submits-license-application-to-nrc-for-the-palisades-twin-smr-300s/
[20] U.S. Department of Energy, Office of Nuclear Energy, “U.S. Department of Energy Reactor Pilot Program.” https://www.energy.gov/ne/us-department-energy-reactor-pilot-program
[21] Idaho National Laboratory, “America Built Nuclear Power in Idaho First. Now It’s Doing It Again” (John Wagner remarks), August 2026. https://inl.gov/feature-story/america-built-nuclear-power-in-idaho-first-now-its-doing-it-again/
[22] NucNet, “Antares Achieves Initial Criticality of Privately Developed Advanced Nuclear Reactor” (Chris Wright remarks), June 5, 2026. https://www.nucnet.org/news/antares-achieves-initial-criticality-of-privately-developed-advanced-nuclear-reactor-6-5-2026
[23] U.S. Department of Energy, “Energy Department Awards $94 Million to American Companies to Help Expedite the Deployments of Small Modular Reactors in the United States,” May 2026. https://www.energy.gov/articles/energy-department-awards-94-million-american-companies-help-expedite-deployments-small
[24] Rigzone, “US DOE Awards Over $94MM to Light Water SMR Advancement Projects” (Chris Wright remarks), May 18, 2026. https://www.rigzone.com/news/us_doe_awards_over_94mm_to_light_water_smr_advancement_projects-18-may-2026-183713-article/
[25] U.S. Department of Energy, “Department of Energy Announces American Nuclear Supply Chain Loans” ($17.5 billion conditional commitment; Chris Wright remarks), June 23, 2026. https://www.energy.gov/articles/department-energy-announces-american-nuclear-supply-chain-loans
[26] S&P Global Commodity Insights, “US DOE to Loan $17.5 Billion to Five Nuclear Reactor Projects for Long-Lead Components,” June 23, 2026. https://www.spglobal.com/energy/en/news-research/latest-news/electric-power/062326-us-doe-to-loan-175-billion-to-five-nuclear-reactor-projects-for-long-lead-components
[27] Office of the Texas Governor, “About TANEO — Texas Advanced Nuclear Energy Office.” https://gov.texas.gov/taneo/page/about-taneo
[28] Brian Martucci, Utility Dive, “Texas Opens $350M Advanced Nuclear Grant Program,” April 2, 2026. https://www.utilitydive.com/news/texas-opens-350m-advanced-nuclear-grant-program/816442/
[29] Office of the Texas Governor, “TANEO News” (statements of Gov. Greg Abbott). https://gov.texas.gov/taneo/page/taneo-news
[30] Office of Governor Gretchen Whitmer, “Gov. Whitmer Launches Michigan Affordable and Responsible Growth Action Plan, Calls on Data Center Companies to Sign Pledge” (including Casey Katims, U.S. Climate Alliance remarks), July 15, 2026. https://www.michigan.gov/whitmer/news/press-releases/2026/07/15/gov-whitmer-data
[31] Kyle Davidson, Michigan Advance, “Whitmer Rolls Out Data Center Plan, Calls on Lawmakers to Pass Laws Protecting Energy Customers” (Gov. Gretchen Whitmer remarks), July 15, 2026. https://michiganadvance.com/2026/07/15/whitmer-rolls-out-data-center-plan-calls-on-lawmakers-to-pass-laws-protecting-energy-customers/
[32] Tennessee Department of Economic and Community Development, “Centrus to Expand Oak Ridge Centrifuge Manufacturing Plant to Facilitate Large-Scale Deployment” (Gov. Bill Lee remarks), January 23, 2026. https://tnecd.com/news/centrus-to-expand-oak-ridge-centrifuge-manufacturing-plant-to-facilitate-large-scale-deployment/
[33] Chattanooga Times Free Press, “Lee Wants Tennessee to Host Nuclear Fuel, Waste ‘Campus’” (with tabulation of Oak Ridge nuclear investments), April 2026. https://tfpdev.wehco.com/news/2026/apr/02/lee-wants-tennessee-to-host-nuclear-fuel-waste/
[34] Office of Governor Abigail Spanberger, Commonwealth of Virginia, “Governor Spanberger Highlights Virginia’s First-of-Its-Kind Data Center Energy Consumption Tax” (POLITICO Energy interview), July 6, 2026. https://www.governor.virginia.gov/newsroom/news-releases/2026/july-releases/name-1120725-en.html
[35] Bradley J. Nowak, Williams Mullen, “Virginia Budget Creates New Electricity Consumption Tax for Data Centers,” June 30, 2026. https://www.williamsmullen.com/insights/news/legal-news/virginia-budget-creates-new-electricity-consumption-tax-data-centers
[36] Office of Governor Abigail Spanberger, Commonwealth of Virginia, “SCC Orders Data Centers to Cover the Cost of Transmission Infrastructure Built Exclusively for Those Facilities” (Josephus Allmond remarks), August 5, 2026. https://www.governor.virginia.gov/newsroom/news-releases/2026/august-releases/name-1122057-en.html
[37] Commonwealth of Pennsylvania, Office of Governor Josh Shapiro, “Gov. Shapiro Releases Full GRID Standards to Protect Pennsylvanians,” May 27, 2026. https://www.pa.gov/governor/newsroom/2026-press-releases/gov-shapiro-releases-full-grid-standards-to-protect-pennsylvania
[38] Commonwealth of Pennsylvania, Office of Governor Josh Shapiro, “PA House Passes Legislation to Codify Gov. Shapiro’s GRID Standards,” June 2026. https://www.pa.gov/governor/newsroom/2026-press-releases/news–pa-house-passes-legislation-to-codify-gov-shapiro-s-grid-s
[39] Spaziocrypto, “Nuclear Power for AI 2026: Microsoft, Google and Meta Act” (interview remarks of Prof. Jacopo Buongiorno, Director, MIT Center for Advanced Nuclear Energy Systems), May 2026. https://en.spaziocrypto.com/data-centers/nuclear-power-ai-data-centers-microsoft-google-meta/
[40] David Roberts, Volts, “For Data Centers, a Little Flexibility Goes a Long Way” (discussion of Princeton ZERO Lab / encoord / Camus study with Prof. Jesse Jenkins, Princeton University), March 25, 2026. https://www.volts.wtf/p/for-data-centers-a-little-flexibility
[41] International Energy Agency, “IEA Executive Director Addresses World Leaders on Crucial AI-Energy Nexus at International Summit Chaired by France and India” (Dr. Fatih Birol remarks, AI Action Summit, Paris). https://www.iea.org/news/iea-executive-director-addresses-world-leaders-on-crucial-ai-energy-nexus-at-international-summit-chaired-by-france-and-india
[42] International Energy Agency, “AI Is Set to Drive Surging Electricity Demand from Data Centres While Offering the Potential to Transform How the Energy Sector Works” (Dr. Fatih Birol remarks on the Energy and AI report). https://www.iea.org/news/ai-is-set-to-drive-surging-electricity-demand-from-data-centres-while-offering-the-potential-to-transform-how-the-energy-sector-works
[43] Tom’s Hardware, “Google, Microsoft, Meta, and Amazon Capex Spending to Hit $725 Billion in 2026, Up 77% From Last Year,” April 30, 2026. https://www.tomshardware.com/tech-industry/big-tech/big-techs-ai-spending-plans-reach-725-billion
[44] CNBC, “Tech AI Spending Approaches $700 Billion in 2026, Cash Taking Big Hit,” February 6, 2026. https://www.cnbc.com/2026/02/06/google-microsoft-meta-amazon-ai-cash.html
[45] The White House, “National Security Presidential Memorandum/NSPM-11: Artificial Intelligence in the National Security Enterprise,” June 5, 2026. https://www.whitehouse.gov/presidential-actions/2026/06/national-security-presidential-memorandum-nspm-11/
[46] U.S. Department of Energy, “U.S. Department of Energy Meets President Trump’s Goal, Delivers Third Advanced Reactor Criticality” (Executive Order 14301 milestones), July 2, 2026. https://www.energy.gov/articles/us-department-energy-meets-president-trumps-goal-delivers-third-advanced-reactor
[47] Brookfield Asset Management, Form 8-K, “The U.S. Department of Energy Announces Conditional $17.5 Billion Financing to Support Westinghouse Nuclear Reactor Deployment,” June 23, 2026. https://www.sec.gov/Archives/edgar/data/0001937926/000117184326004262/exh_991.htm
[48] The Korea Times, “HD Hyundai, SK Join Bill Gates’ TerraPower for SMR Biz,” August 14, 2026. https://www.koreatimes.co.kr/business/companies/20260814/hd-hyundai-sk-join-bill-gates-terrapower-for-smr-biz
[49] Yahoo Finance / Motley Fool, “X-Energy Q2 Earnings Call Highlights” (Q2 FY2026 results: revenue +154%, $1.9B cash, HALEU agreements), August 14, 2026. https://finance.yahoo.com/energy/articles/x-energy-q2-earnings-call-190421323.html
[50] World Nuclear News, “Criticality for Fourth US Microreactor to Meet 4 July Deadline” (Aalo Atomics Critical Test Reactor), July 7, 2026. https://www.world-nuclear-news.org/articles/criticality-for-fourth-us-microreactor-meets-deadline
[51] ESG Dive, “Meta Inks Trio of Nuclear Deals to Support AI Ambitions” (Chris Levesque, TerraPower, remarks), January 9, 2026. https://www.esgdive.com/news/meta-inks-nuclear-deals-terrapower-oklo-vistra-support-ai-ambitions-rfp-6-6-gw/809231/
[52] WBIW, “Gov. Braun Launches Advanced Nuclear Ready Community Program” (Sec. Suzanne Jaworowski remarks), July 15, 2026. https://www.wbiw.com/2026/07/15/gov-braun-launches-advanced-nuclear-ready-community-program/
