Data on the current costs of small modular nuclear reactors (SMR) is starting to roll in. As a result, it’s now possible to make some projections of how long it would take for their costs to drop to the level of renewables today. The results aren’t good for SMRs.
Who are our contestants? NuScale and Long Energy have signed contracts, definite in the case of the former and of unknown quality in the case of the latter. As a result, we know their current costs, long before they actually deliver any electricity anywhere. I’m going to give them massive benefits of the doubt that their costs won’t multiply further, a very conservative concession given that they are both first of a kind technologies which have no working, deployed units yet.
Both are American firms. Per crunchbase, one of my go-to sources for funding insights, NuScale has raised about $470 million and is based in Oregon. Per LinkedIn, they have about 570 employees and are in the oil and gas sector. I assume the first is mostly accurate, but that the second isn’t. Last Energy is based in Washington, DC, has raised $3 million, and has about 40 employees.
What are they claiming in terms of contracts and deliveries? Well, NuScale is way behind schedule and over budget on delivering six 77 MW units to an Idaho National Laboratory site. Even with direct government funding of $1.4 billion from the US DOE and a $30 per MWh tax break from the Inflation Reduction Act, it’s still coming in at $89 per MWh wholesale cost of electricity. Unsubsidized, lets generously call it $120 per MWh, approaching the average retail price of electricity in the US.
As for Last Energy, with its $3 million in funding, Beltway location, and <50 employees, it recently announced deals in Poland and the UK for 34 reactors, each 20 MW in capacity, for an eye watering $19 billion. That works out to about $161 per MWh wholesale at best, above the retail price of electricity in those countries, and remarkably even above the cost per MWh of the far behind schedule and far over budget Hinkley Point C project. Let’s pretend that a firm with $3 million in funding and <50 employees have signed deals worth $19 billion and we should believe these deals are remotely firm. We are, after all, giving SMRs every benefit of the doubt in order to see what might happen.
The premise of small modular reactors is that the observed reality of economies of scale due to mass manufacturing will kick in. It was first observed in the 1930s by an efficiency expert, and is often referred to as Wright’s Law after him. The Boston Consulting Group stole the idea, called it the experience curve and sold it to their clients along with a strong recommendation to create monopolies with it. Sometimes it’s called the learning curve.
At heart, all it says is that manufacturing experience is an s-curve of cheaper costs per unit. Costs stay level for a bit at the beginning, then drop by 20% to 27% with every doubling of numbers, and then after a bunch of doublings flatten out again. Stuff like screws, nuts, and toaster elements are as cheap as they are going to get because we’ve made millions of them and they are in flat part of the curve at the end. Things like small modular reactors are still in the flat part of the curve at the beginning.
I put a question to Professor Bent Flyvbjerg, a global expert on modularity and megaprojects, a year or so ago about SMRs and Wright’s Law. He’d asked if he could include some of my material on the natural experiment of wind and solar vs nuclear in China in his book, How Big Things Get Done, which was published a month or so ago. That gave me the opportunity to ask his deeply informed opinion on the subject. The question I asked was about the number of units in the initial flat part of the s-curve. He said dozens. I took that to mean perhaps 60 or 70 units before the doubling cost reduction really kicked in. Someone commented on a post of mine on LinkedIn that Rolls Royce expected it would have to manufacture and sell 50 of its proposed units before prices fell, but I haven’t validated that statement, and as its SMRs are 470 MW capacity that’s about 24 GW of sales regardless.
So I had a starting point for two SMR technologies, a doubling ratio, and could create a couple of scenarios, one for each. Doubling would have different effects, since NuScale was starting at a lower cost point per MWh — and remember I’m giving both of them the huge benefit of the doubt that either will come in at current costs — but had units about 4 times the capacity, so doubling of volume would be slower. Last Energy purports to have tiny units and is asserting that it has deals for a lot more of them numerically, but is starting at a much higher cost point. Once again, to give SMRs every benefit of the doubt, I’m going to pick a high learning curve value of 25%, not halfway between 20% and 27%.
I decided to go with an aggressive sales profile for both companies through 2040, with both experiencing massive successes in selling more of their very expensive, unproven, first of a kind products. Like assuming that their current costs wouldn’t rise radically, this was once again a very conservative option that was very much in their favor in terms of the analysis.
Selling lots more units every couple of years through 2040 would lead to these companies which have delivered nothing so far having sold about 500 units with a capacity around 40 GW for NuScale and about 2,500 units with a capacity of around 50 GW for Last Energy.
For context, there are only about 440 operating commercial nuclear reactors in the world with a combined capacity of about 400 GW, so these numbers would represent a massive increase in the number of reactors and 10% and 13% of global nuclear generation capacity increases in 17 years. Once again, this is an absurdly optimistic forecast for this technology, and is incredibly favorable to their price decreases.
And what are the results?
Well, with massive increases in numbers in both cases and with the assumption that they’ll hit current cost projections, neither gets close to the current global averages for wind or solar. Neither gets below $50 per MWh in 2023 dollars. Coincidentally, both get to $51 per MWh, which wasn’t something I gamed, just something that fell out the bottom of my really aggressive success scenarios for both of them.
When I did the analysis of the SMR space initially about two years ago, I found that there were about 18 designs extant at the time. They are all competing with one another. They have different governmental backers. As I pointed out in discussions of the space, the only chance any has of becoming a cheap form of generation is if a major geographical region like the US or EU picks a winning design, forces it down everyone’s throats, and as a result maybe reaps the benefits of Wright’s Law.
But these two scenarios would already add 23% to global nuclear capacity when it’s been pretty flat for a couple of decades. And nine times as many firms would double current nuclear capacity. That’s really unlikely when all of this SMR electricity would be so expensive and wind, solar, transmission, and storage are so cheap.
What’s more likely to happen is pretty obvious. The odds that Last Energy delivers anything approaches zero, and its deals are mostly likely of the same quality as a lot of SPAC MOUs and LOIs that have riddled the space. It wouldn’t surprise me if they are looking for a SPAC reverse takeover right now. As a result, the company getting to 2,500 delivered units and achieving not terribly expensive energy in 2040 (when we need lots of it by 2030) is something I would find extraordinarily surprising.
As for NuScale, municipal partners in the deal keep leaving it, costs keep rising, the first connection to the grid has been pushed back to 2029 with the current unrealistic schedule, and it has MOUs and design contracts in a few places around the world with the faint hope that it will deliver something somewhere before 2030. Most of its target customers in the US and elsewhere don’t have nuclear generation in their portfolios today and most of the countries aren’t integrated into the IAEA for commercial generation. The seven overlapping rings of security requirements for commercial nuclear generation I describe in my assessment space for SMRs have only started to be put in place, and they are non-trivial. The odds that NuScale realizes deployment of even 50 of its reactors are low, so the odds that it will get the first increment of value from Wright’s Law approaches zero. They’ll remain expensive, if indeed they ever get grid connections anywhere.
Wright’s Law isn’t going to save the deep inefficiencies of SMRs. As I pointed out two years ago, the world tried tiny commercial nuclear reactors in the 1960s and 1970s, they were too expensive due to the physics of thermal generation, and SMRs wouldn’t be successful in overcoming that with massive numbers of units.
I don’t like paywalls. You don’t like paywalls. Who likes paywalls? Here at CleanTechnica, we implemented a limited paywall for a while, but it always felt wrong — and it was always tough to decide what we should put behind there. In theory, your most exclusive and best content goes behind a paywall. But then fewer people read it! We just don’t like paywalls, and so we’ve decided to ditch ours.
Unfortunately, the media business is still a tough, cut-throat business with tiny margins. It’s a never-ending Olympic challenge to stay above water or even perhaps — gasp — grow. So …