Why nuclear is back: AI data centers and the SMR moment

The AI buildout has done what twenty years of climate policy could not — make new nuclear urgently economic. AI training and inference demand vast quantities of firm, 24/7 power, and the grid cannot deliver it on the timeline hyperscalers need. That has pushed operators toward generation they can secure directly, and nuclear — both plant restarts and a new generation of small modular reactors (SMRs) — has moved from speculation into active construction. Hyperscalers are now signing long-term power purchase agreements with nuclear developers, and reactor projects that sat dormant for a decade are breaking ground.

The signal moment came in 2026. The U.S. Nuclear Regulatory Commission issued a construction permit for TerraPower's Natrium plant in Wyoming — the first commercial reactor the agency had approved for construction in nearly a decade, and the first non-light-water commercial design in more than 40 years.[1] Weeks later, the company mobilized to break ground. For a sector that had not started a genuinely new reactor type in two generations, it was a structural turning point.

But the buildout has run ahead of the workforce that can deliver it. The U.S. is rebuilding nuclear-construction capability that largely atrophied after the late 1970s, on first-of-a-kind designs, under the most demanding documentation standard in construction. That gap — not capital, not licensing — is the binding constraint on the restart, and it is what this guide is about. Start with our flagship coverage of SMR-powered data center developments, the DOE-aligned momentum in Clayco's nuclear data center campus proposal, and how operators are evaluating colocation with nuclear storage. For the broader context, see the Data Center Construction guide.

The Vogtle precedent: what the last build taught us

Any honest account of America's nuclear restart begins at Plant Vogtle in Georgia. Units 3 and 4 were the only new commercial reactors built in the U.S. in more than 30 years, and they became a cautionary tale: at completion, the project's final cost of roughly $36 billion and its roughly 15-year schedule were both more than double the original estimates.[4] The reasons matter, because every SMR and advanced-reactor team building today is explicitly trying not to repeat them.

The Vogtle overruns were not, at root, a technology problem. They traced to a workforce and execution problem: a domestic nuclear-construction industry that had lost its muscle memory, design changes that rippled through an unprepared supply chain, and quality-documentation rework that conventional construction crews were not equipped for. The project ultimately employed more than 7,000 construction workers at peak — and the difficulty of assembling, training and retaining a nuclear-qualified workforce of that size was a central driver of the delays.[4]

There is a more hopeful reading, and it is the thesis behind the current wave. Industry veterans describe nuclear cost as a learning curve: the first unit of any design is always the most expensive and the slowest, and costs fall sharply with repetition. China, which is building more than 30 reactors and never let its construction base atrophy, delivers comparable plants at a fraction of Vogtle's cost.[5] The bet embedded in SMRs is that factory-built, repeatable, smaller units can climb that learning curve far faster than one-off gigawatt megaprojects — if the workforce exists to build the first ones well. That conditional is the whole game.

Active SMR & advanced reactor projects

Multiple first-of-a-kind builds are now in construction or formal NRC review in the U.S. These are the projects setting the pace — and competing for the same scarce pool of nuclear-qualified construction talent. The list below reflects the most active programs; the pipeline behind them is deeper and moving quarter to quarter.

TerraPower officially started construction in April 2026, mobilizing roughly 1,600 workers over the build, after the NRC's March 2026 construction permit — and notably cleared its NRC safety review ahead of schedule and under budget.[2] NANO Nuclear's KRONOS MMR reached a major regulatory milestone when the NRC formally accepted its construction permit application in May 2026, making it the first commercially-ready microreactor to reach that stage.[3] Track the wider buildout in the Data Center News hub and our continuing SMR developments coverage.

What makes nuclear construction different

It is tempting to treat a reactor as just another complex industrial build. It is not, and the difference is precisely what makes the workforce so hard to assemble. Three things set nuclear construction apart from even the most demanding commercial or data center work.

The documentation standard is total

Nuclear construction is governed by NQA-1, the ASME nuclear quality-assurance standard, layered with NRC oversight and ASME Section III for nuclear components. In practice this means every material, weld, pour and inspection is traceable, verified and documented to a standard no other construction sector approaches. A crew that is excellent at commercial concrete or industrial piping is not automatically qualified to do the same work on a nuclear island, because the documentation and verification discipline — not just the physical skill — is the deliverable.

There is effectively no margin for rework

On a commercial build, a nonconformance is a punch-list item. On a nuclear build, a single undocumented deviation can halt work, trigger regulatory review, and ripple into months of delay — exactly the dynamic that compounded at Vogtle. This raises the premium on getting it right the first time, which in turn raises the premium on experienced supervision: people who have actually delivered nuclear scope and know where the traps are.

The regulatory interface is continuous

NRC interaction is not a permitting gate you clear once; it is a continuous relationship throughout construction. Project leaders who understand how to work with the regulator — how to document, when to notify, how to manage change without triggering re-review — materially shape the schedule. That experience is rare, and it is concentrated in a small, aging pool of professionals.

The construction workforce gap

The deeper problem is generational. The U.S. has not built commercial reactors at scale since the 1970s and 1980s, so the population of supervisors, QA/QC inspectors, nuclear welders and senior operators who have personally delivered a reactor has thinned to a fraction of what the current pipeline requires. Vogtle absorbed much of what remained; the SMR wave now arriving has to rebuild that capability largely from scratch, and in parallel across multiple states at once.

This is why nuclear-qualified talent — not capital, and increasingly not even licensing — is the binding constraint on the restart. A developer can raise money and clear the NRC and still stall for lack of people who can execute to NQA-1. See the broader picture in the nuclear power talent shortage, the practical realities in nuclear construction hiring challenges, and the playbooks emerging in staffing strategies for large-scale nuclear infrastructure and staffing first-of-a-kind nuclear projects.

Roles in highest demand

The hiring pressure concentrates in a small set of titles where nuclear experience is non-negotiable. None of these transfer cleanly from commercial construction, and all are booked far ahead of need.

Nuclear construction project managers

PMs who have delivered nuclear scope and can manage the NRC interface, the change-control discipline and the documentation burden. This is the single hardest role to fill and the one whose absence most reliably derails a schedule.

QA/QC leads fluent in NQA-1

Quality leads who live in the NQA-1 and ASME Section III world. They are the backbone of a nuclear project's no-rework imperative, and the supply is thin because the credential plus real reactor experience is rare.

Superintendents with reactor or large-scale energy experience

Field leaders who understand that nuclear trade coordination is denser and the construction-to-commissioning handoff happens earlier than on commercial work.

SMR project-delivery leadership

A newer category: leaders who can run a program of repeatable modular units rather than a single bespoke build — the people who will determine whether the SMR learning-curve thesis actually pays off.

• Role demand
  In-demand nuclear construction rolesThe titles gating the schedule
• Leadership
  Specialized construction leadership for nuclearWhat sets nuclear leaders apart
• Team building
  SMR developers building delivery teamsHow developers are sourcing it

Explore the core roles directly — construction project manager, superintendent, and QA/QC manager.

Certifications & qualifications that matter

Nuclear construction layers federal regulatory qualifications over the standard construction credentialing stack — and it is the combination that's scarce. Plenty of professionals hold a PMP or a trade license; far fewer pair it with documented nuclear-quality experience. The qualifications that move a candidate to the top of the list:

• NQA-1 experience — demonstrated work to the nuclear quality-assurance standard; the single most differentiating line on a nuclear construction résumé.
• NRC licensing familiarity — understanding of the Part 50 / Part 52 process and how to manage construction within it.
• ASME Section III — for the welding, fabrication and QA disciplines tied to nuclear-grade components.
• Senior Reactor Operator (SRO) pipeline — on the operations side, the NRC-licensed operator credential is a recognized long-lead constraint that owners must plan years ahead.

For the broader credentialing landscape (PMP, BCxP, NETA, NICET and the rest of the stack), see the Construction Certifications guide.