For the first time since the agency was created in 2000, the National Nuclear Security Administration (NNSA) has furloughed about 1,400 federal employees, roughly 80% of its workforce, because of the October shutdown. About 375 employees remain on duty for essential functions.
That’s a break with the 35-day 2018–2019 shutdown, when NNSA avoided furloughs. Sen. Ed Markey underscored the contrast in a letter to DOE, calling the current furloughs “completely avoidable.”
Why this matters for nuclear energy
Disruptions on the defense side ripple into the civilian buildout. The U.S. defense nuclear enterprise props up the people, vendors, modeling tools and fuel-cycle capacity that next-gen reactors need; when federal stewards step away, oversight pauses, reviews stack up, and schedules slip.
Defense-side research matters acutely for civilian deployment. Los Alamos’s MCNP radiation-transport code, developed inside the weapons complex, is a workhorse across reactor physics, shielding, and safety analysis in industry. DoD’s Project Pele stands up domestic TRISO/HALEU fuel manufacturing that directly overlaps with civil microreactor needs, and DOE’s HALEU Availability Program has already started U.S. production at Centrus, the same feedstock many advanced reactors require.
The furloughs are having an immediate impact on programs. For instance, DOE’s Reactor Pilot Program, launched in June and formally naming 11 projects in August with a target to reach criticality in at least three test reactors by July 4, 2026, now faces schedule risk because federal overseers are off the job. Companies in the initial cohort include Oklo, Last Energy, Aalo Atomics, Terrestrial Energy, and others.
In parallel, DOE re-opened its $900 million Generation III+ SMR solicitation in March to support near-term light-water SMR deployments—work that depends on steady federal engagement.
University partnerships are also navigating the policy whiplash from the furloughs: DOE’s attempted 15% cap on indirect costs was blocked in federal court, but uncertainty has already complicated planning. Source.
U.S. electricity use is already at record levels and still rising in 2025–2026, driven in large part by data centers and AI. The EIA and multiple analysts project brisk demand growth; Grid Strategies estimates tens of gigawatts of incremental data-center load by decade’s end.
Goldman Sachs estimates global data-center power demand could rise 165% by 2030; one GS brief adds that supplying all incremental data-center growth with nuclear would imply about 85–90 GW of new nuclear capacity.
RAND’s modeling stretches the bounds of what’s coming: in aggressive scenarios, single AI training runs could demand up to 1 GW by 2028 and as high as 8 GW by 2030 at a single location.
The grid backdrop is strained. NERC’s latest Long-Term Reliability Assessment warns that well over half of North America faces elevated or high risk of energy shortfalls over the next 5–10 years as demand accelerates and retirements mount.
In September, FERC Chair David Rosner told grid operators, “We cannot efficiently plan the electric generation and transmission needed to serve new customers if we don’t forecast how much energy they will need as accurately as possible,” in a public letter on large-load forecasting, as FERC notes on its website.
What industry is doing anyway
Even amid the shutdown, tech firms are homing in on nuclear. Microsoft and Constellation signed a 20-year deal tied to restarting Three Mile Island Unit 1, with work budgeted around $1.6B and operations targeted for 2027–2028.
Google inked a first-of-its-kind agreement with Kairos Power to deploy about 500 MW of advanced reactors by 2035 (first unit by 2030), including a TVA-backed Hermes 2 project in Oak Ridge. Source.
Amazon has taken a two-track approach: it bought a nuclear-adjacent, 960-MW data-center campus at the Susquehanna plant for $650M, and separately unveiled plans with X-energy/Energy Northwest for up to 12 SMRs (~960 MW) in Washington state.
DOE says nuclear is essential for reliability and for competing in AI, yet this shutdown imposed the first-ever NNSA furloughs, introducing avoidable delay risk to programs meant to expand firm, clean power. That contradiction is the story: we’re pausing oversight and testing just as the grid and the AI economy need more dispatchable, carbon-free capacity, not less.
Read the full article from the Source
