SMRs in ~1k Words
Bringing nukes from monumental project to iterable product
👋Hey!
Happy to be back for another round of my Tuesday posts! I enjoyed reading a few “___ in under 1,000 words” posts this past week that I thought did a good job of giving some baseline familiarity to new topics. In my opinion, these fit well with the whole Uncredentialed thesis, so I decided to give it a shot.
Turns out, I could work on my brevity. I didn’t cut all the way down to 1k, but this still will serve as my shortest primer/deep dive by a longshot so eager to hear thoughts on this compared to ones that are more medium-length:
and longform:
I think I have a preferred pick out of those 3 but I don’t want to poison the well, so just let me know what you prefer!
In other news, as an update from last week, the replacement part for my coffee grinder came early, but didn’t fix the issue. There’s a very thin line of responsibility between me and a Costco pack of Monster energy drinks and it’s getting thinner by the day.
Now that we’re already on the topic of (probably) hazardous chemicals, let’s talk nukes!
But first! If this is your first time here, welcome to Uncredentialed! Here at Uncredentialed, we’re all about agency over permission, taking action, learning, and iterating, regardless of having the right education, experience, or investments. I write 2x weekly about tech, startups, and strategy and would love to have you join the community. Sign up below!
SMRs in ~1k Words
If you followed my data center series, you know about the mismatched silicon and grid clocks. We have chips sitting in inventory because we don’t have enough power to run them. The grid is old, brittle, and wasn’t built for 24/7 data center loads.
Small modular reactors (SMRs) keep showing up in the headlines because they promise a potential solution to this mismatch, but most coverage on them either overloads on jargon or handwaves the physics entirely.
Here’s my attempt to explain what SMRs are, how they differ from traditional nuclear, and who’s racing to build them in an approachable way that can serve as a launch point for learning more about the industry.
How Nuclear Power Actually Works
At its roots, nuclear power is just a fancy way to boil water.
Inside a reactor, you have fuel rods made of uranium. When a neutron hits a uranium atom, the atom splits (this is fission), releasing energy and more neutrons. Those neutrons go on to split more atoms, creating a chain reaction, with the energy from all that splitting being released as heat.
Water is then run by to absorb the heat, where it turns into steam, that steam spins a turbine, and the turbine generates electricity. It’s the same basic principle as a nat gas or coal plant, just with nuclear fission generating the heat instead of combustion.
The process has 3 major engineering challenges:
Controlling the chain reaction: This is solved by control rods that absorb neutrons to slow things down. Without them, fast-hurtling neutrons would increase exponentially, creating an out-of-control reaction.
Efficient heat removal: If the water isn’t absorbing heat fast enough or the pumps stop pushing new water through the system, heat will grow and things can melt down.
Radioactive containment: This one’s probably obvious. We need to be smart about how we handle and dispose of radioactive elements. They don’t just go out in the dumpster once spent.
Traditional nuclear plants solved these problems with size and redundancy. Big containment domes, multiple backup cooling systems with pumps and diesel generators, and more worked together to provide safe, reliable power, but at a cost of $15B+ and 10-15 years of construction.
Why Traditional Nuclear Got So Expensive
Nuclear economics traditionally favored big reactors. A 1k MW reactor produces 10x the electricity of a 100 MW reactor but doesn’t require 10x the security staff, licensing work, or containment structure. Bigger reactors spread fixed costs over more electrons.
Size brought its own problems however, turning every new traditional plant into a bespoke megaproject. The reactor vessel, containment building, and cooling systems are all designed and built on-site, according to that site’s specific requirements. There’s no standardization or iteration, which limits the development of learning curves, and means your, say, 50th plant will likely cost just as much as your 1st (and maybe even more if regulation continues to grow higher)!
This is the opposite of how complex technology should be built. As one example, SpaceX famously disrupted the lethargic rocket industry by iterating rapidly, reusing hardware, and treating rockets like products to be manufactured, rather than monuments to be constructed. Traditional nuclear hasn’t seen that benefit because we never built enough standardized units to learn and iterate.
The SMR Difference
I’d break the benefits of SMRs down into 4 categories.
Learning Curves: Instead of building reactors on-site, you manufacture them in factories, before shipping them to location. The reactor core arrives on a truck, gets installed, and connects to the grid. By constructing the SMRs in scaled factory settings, teams will be able to engage in process improvement and make the 50th plant cost exponentially less than the 1st.
Smaller Upfront Bets: A $1B-$2B SMR project carries different risk than a $15B megaproject, especially once consistent, high quality manufacturing has been proven. Customers can simply add more capacity incrementally as needed and trust is built.
Passive Safety Systems: Most SMR designs rely on physics rather than mechanical systems to prevent meltdown. If power fails, convection and gravity naturally circulate the coolant without backup pumps.
Alternative Coolants: Traditional reactors almost all use pressurized water. Many SMRs experiment with other methods like liquid sodium, molten salt, or helium. These alternatives can operate at higher temps (more efficient) or lower pressures (simpler containment).
It’s not all sunshine and roses. SMRs have their skeptics and not without reason. Nobody’s proven factory economics at scale yet, for now it’s just a nice story.
Further, NuScale, the first company to get an SMR design certified in the US, cancelled their flagship project in late 2023 after cost projections grew from $3.6B for a 720 MW plant to $9.3B for a 462 MW plant.
Now, in all fairness, the largest driver of that cost increase came from financing costs on long term debt. The Fed Funds Rate rising from near-0% to over 5% will do that to you. The next biggest cost increase was from the inflation that preceded the rate hikes. Critics would disagree, but I’m willing to bet on the other side, that the learning curves built from manufacturing modular reactors at scale will not only be enough to offset these risks, but should actually reduce financing risk over time as costs come down.
The DOE’s Reactor Pilot Program
In 2025, the DOE launched its Reactor Pilot Program aimed at getting at least 3 test reactors to criticality by July 4, 2026 (America’s 250th). They selected 10 companies:
Valar Atomics: Helium-cooled, high temperature gas reactor using TRISO fuel
Radiant: 1 MW helium-cooled microreactor, among the first to test at Idaho Nation Lab’s (INL) new facility
Oklo: 75 MW liquid metal cooled fast reactor, first unit under construction at INL
Atomic Alchemy (Oklo subsidiary): 15 MW light water reactor for radioisotope production
Aalo Atomics: 10 MW sodium-cooled reactor, with a 50 MW version designed for data centers
Antares Nuclear: 500 kW sodium heat pipe microreactor for military installations
Last Energy: 20 MW pressurized water reactor, planning 30 units at a single Texas site
Natura Resources: Molten salt reactor, first of its type to get an NRC construction permit
Deep Fission: 15 MW reactor designed to be built one mile underground via a 30-inch borehole
Terrestrial Energy: 195 MW molten salt reactor, targeting commercial operation within 5 years
Progress quickly took off after the announcement as capital flowed into the space and timelines condensed. The program’s done a good job of bringing in a wide range of approaches with reactors ranging from 500 kW to 195 MW, all sorts of coolants, and target customers ranging from the grid, to data centers, to military bases, and even to Mars. I have confidence that running such a wide array of parallel experiments will be able to find the right solution for commercializing SMRs.
Where this is heading
The Reactor Pilot Program feels like a good step in the US’ journey towards a better balance between engineering and lawyering. As a Gen-Zer, I’ve mostly only known a future built in the world of bits, not the world of atoms, but this should help fix that.
What makes me more optimistic than any specific company or reactor design is the fact that we’re running a portfolio of real experiments. Different coolants, sizes, customers, and deployment models will compete in parallel, and the winners will emerge through iteration, not theory.
If even a couple of these designs reach commercial scale, they could reshape how we think about building our energy infrastructure altogether. The next couple of years should tell us whether SMRs are able to pivot nuclear energy from a monumental project to an iterable product, and if they can, the world of atoms will begin to look radically different.
Thanks for reading Uncredentialed! If you can think of anyone else who’d find this interesting, send it their way! If you want to discuss more, please reach out! And, lastly, if you haven’t already, subscribe to Uncredentialed!