RegulationConcernsGetting an NRC design certification used to be the finish line for a reactor company. Now it’s barely the starting gun. 🏗️ NuScale learned that the hard way when its Carbon Free Power Project collapsed in November 2023 despite being the only US-certified SMR design on the market. Certification didn’t save it. Cost did. Five real obstacles stand between today’s SMR industry and the fleet-scale future everyone keeps promising, and the companies actually solving them are worth naming specifically, not vaguely praised.
First-of-a-kind costs that eat the business case
The NuScale/UAMPS story is the industry’s cautionary tale, and it’s worth getting the numbers right. The Carbon Free Power Project started in 2021 with a target price of $58/MWh and a $5.3 billion construction estimate for six 77-MWe modules. By 2023, that had climbed to $89/MWh and $9.3 billion, roughly $20,000 per kilowatt, which put it in the same range as the very large reactor projects SMRs were supposed to undercut. 💸 Utah Associated Municipal Power Systems, a coalition of 50-plus small municipal utilities, couldn’t get enough of its members to sign subscriptions at the new price, and the project died.
2021 estimate: $58/MWh, $5.3 billion total
2023 estimate: $89/MWh, $9.3 billion total
DOE had committed up to $1.4 billion in cost-share funding
Industry estimates suggest 5 to 7 deployed units before costs reach Nth-of-a-kind territory
The lesson the rest of the industry took from this, correctly I think, is that customer structure matters as much as the reactor design. Meta has committed up to 6.6 GW across TerraPower, Oklo, Vistra, and Constellation. Google locked in 500 MW from Kairos Power. Amazon put $700 million into X-energy. These are companies that can absorb a first-of-a-kind cost overrun without a subscription vote, and SMR Intel’s own analysis of the NuScale collapse is blunt about why that structural difference matters more than any single technical achievement. On the government side, the DOE Loan Programs Office has kept working its Title 17 financing muscle, disbursing part of a $1.52 billion conditional guarantee toward restarting the Palisades plant in Michigan, alongside a re-announced Gen III+ SMR program offering $800 million for up to two projects.
Have you noticed how every surviving SMR deal now leads with the name of a deep-pocketed anchor customer before it mentions the reactor design? That’s not a coincidence. It’s the whole industry adjusting to what NuScale taught it.
Fuel that barely exists yet
Most advanced reactor designs run on HALEU, uranium enriched between 5% and 20%, and until recently the only commercial supplier at scale was Russia’s Tenex, which the US banned in May 2024. ⚛️ That left a real supply gap. Centrus Energy has been racing to close it at its Piketon, Ohio facility, producing more than 920 kilograms of HALEU from a demonstration cascade between October 2023 and mid-2025. In January 2026, the Department of Energy signed Centrus to a $900 million task order to move from demonstration to commercial-scale production, part of a broader $2.7 billion, ten-year DOE commitment to expand domestic enrichment.
Centrus completed its final demonstration-contract HALEU delivery in mid-2026
Oklo and Centrus announced a planned joint venture in March 2026 to secure Oklo’s own supply
Radiant signed a direct HALEU purchase contract with Centrus for its Kaleidos microreactors, including prepayments to help fund new enrichment capacity
Radiant’s chief nuclear officer, Rita Baranwal, put the company’s logic simply: fuel gets built in parallel with the reactor, not after it. That’s a small sentence carrying a big strategic shift. Companies used to treat fuel availability as someone else’s problem. Now it’s a line item in the reactor development budget itself. 🔬
A workforce old enough to retire before the reactors are built
This challenge gets less attention than cost or fuel, and it probably shouldn’t. Roughly 40% of the existing nuclear workforce is expected to retire within the next decade, and the broader US construction and manufacturing sector was short about 600,000 workers as of April 2025, a gap that’s pushed craft wages up 20% since the pandemic. 👷 Nuclear engineering enrollment fell about 25% between 2012 and 2022 before finally reversing. That’s not a great pipeline to be counting on right as a dozen companies try to build first-of-a-kind reactors simultaneously.
NEI launched NuclearWorks, a career-matching site aimed at students and career-changers
Purdue launched an SMR-focused Nuclear Engineering Professional Credentials program in spring 2026, built as stackable modules rather than a full degree
The University of Michigan reports undergraduate nuclear enrollment up 25% over five years
DOE’s Savannah River Site apprenticeship program now spans 27 occupations
If you’re tracking the nuclear market professionally, SMRbrief Pro turns the nuclear intelligence in this article into a searchable, filterable, always-updated resource you can act on, which is genuinely useful when workforce announcements like these show up scattered across a dozen unrelated press releases. The honest read here is that credential programs and apprenticeships are the right long-cycle answer, but they won’t fully close a gap this size before 2030. Anyone promising otherwise is selling something.
A licensing process built for gigawatt plants, not shipping containers
The NRC’s traditional licensing pathways were designed around single, site-specific gigawatt-scale plants reviewed one at a time, which makes sense for Vogtle and makes no sense for a company trying to deploy fifteen identical 1-MWe units. 📜 The agency has been rewriting its own rulebook to catch up. Part 53, a technology-inclusive framework for advanced reactors, took effect in April 2026. Weeks later, the NRC proposed Part 57 specifically for microreactors of roughly 100 MWe or smaller, which would allow fleet-wide approval of identical designs and fuel loading at the manufacturing facility rather than only on-site.
Part 53: technology-inclusive framework, effective April 29, 2026
Part 57: proposed April 24, 2026, targets fleet licensing for microreactors
Both trace back to the 2024 ADVANCE Act, which directed the NRC to speed up reviews and cut fees
The Federal Register notice describes a goal of issuing operator licenses within six months to a year of application
This is genuinely the most encouraging fix on this list, but it’s still a proposal working through public comment, not a finished tool anyone has used yet. I’d watch the final rule closely before assuming it delivers the speed the industry is promising investors.
Nowhere permanent to put the waste
The United States still has no operating permanent repository for spent nuclear fuel. Yucca Mountain stalled in 2008, and consolidated interim storage keeps getting tied up in litigation. ♻️ Every new SMR built adds to a waste stockpile with no fixed final destination, which is a real problem even though the volumes involved are modest and, according to Argonne National Laboratory research, roughly comparable per unit of energy to a conventional reactor.
Deep Isolation is one of the more concrete answers taking shape. The company launched a full-scale, at-depth deep borehole demonstration program in January 2026 at a facility near Cameron, Texas, working with Halliburton, Amentum, and NAC International, using directional drilling technology borrowed straight from the oil and gas industry to place waste more than a thousand meters underground. In March 2026, Deep Isolation and Oklo published joint results suggesting the same borehole approach could handle waste from recycled advanced-reactor fuel specifically, not just legacy spent fuel.
No permanent US repository exists as of 2026
Deep Isolation’s Cameron, Texas demonstration launched January 2026
The company also holds an ARPA-E SCALEUP award to field-test its Universal Canister System
Finland’s Onkalo repository, opening this year, remains the only operational example worldwide
None of these five problems gets solved by a single company or a single piece of legislation. What’s changed is that each one now has a named entity actually working the problem instead of a slide with the word “someday” on it. Which of these five bottlenecks do you think breaks first, the fuel supply, the workforce, or the waste question?