The narrative around nuclear energy as a tool in the climate counterstrike has changed markedly over the last couple of years, with another recent acceleration following the invasion of Ukraine. I’ve covered nuclear, both fusion and fission, in various Notes over the past year. Here, I’m going to attempt to give the broader context for nuclear power and what contribution we might expect from it in the years ahead. I have confined the discussion to fission here and not included fusion. This ended up being a bit of a brain dump and ran longer than usual, so I’ve included the TL;DR version at the top.
Advantages of nuclear energy:
Base-load, plus can be turned up and down (if not completely on / off), allows it to compensate for wind / solar variability
Efficient use of land (1/300th of solar)
Efficient use of materials
Advantages will be particularly pronounced when we get to higher penetration of renewables, so the greatest value will be far ahead, which compensates in a way for long lead times
Disadvantages of nuclear energy:
Very poor track record of delivering projects outside of China in recent decades, cost overruns normally multiples, not percentages, above original estimates
Long approval and construction times for existing designs
Long development times for new reactors - first SMRs might be online at the end of the decade
While waiting for SMRs to start on the learning curve, the wind and solar continue to get cheaper, the industrial base continues to expand and there is a huge ecosystem of innovation actively working to unplug bottlenecks.
Conclusion (or, shall we say, “working hypothesis”):
Given its advantages, nuclear has a lot to offer the world in terms of scalable, carbon-free heat and power that doesn’t leave the earth blanketed in solar panels. But unless we get a paradigm shift in how it is deployed, it is unlikely to realise its potential. My hope is that the urgency of the climate crisis will enable that paradigm shift to occur, because we need all the tools at our disposal.
Zero Carbon Energy demand: The first thing to say is that the ramp up in zero-carbon energy that we need is formidable, daunting even. We are still a long way from decarbonising even electricity supply and way further still from decarbonising energy supply as a whole. We can’t afford to be overly selective about what technologies we pursue. Whilst we should carefully guard against the pursuit of inefficient pathways in displacement of more efficient pathways. But where it isn’t a case of either / or, then we must pursue both. [Note in the below that the 84% of energy coming from fossil fuels is in primary energy use, i.e. the chemical energy contained within the fossil fuels used, which is a very different thing from final energy use, which is the amount of energy we actually use. Most of the energy in fossil fuels is wasted as heat, so switching to renewables and nuclear, doesn’t require a 1:1 replacement of that energy.]
Here is another great visualisation of the energy mix, this for Europe, from Thunder Said Energy. You can see that the only place where zero-carbon sources have made a dent is in the ‘electricity’ column. That leaves a lot of territory to wrest from fossil fuel’s grasp.
Another fact to contextualise the demand for clean electricity for deep decarbonisation: to replace today’s use of hydrogen (in refining and chemical production), which is almost entirely generated from fossil fuels and emits 1GT / year, with green hydrogen from electricity would use up slightly more than all of today’s wind and solar production. That is before we have started applying hydrogen for any other uses, and there are plenty of ambitions to expand its use.
Nuclear has some standout advantages:
It isn’t intermittent. Capacity factors normally run at 80-90% although (as France discovered this year) they can go down for long periods for maintenance. High capacity factors mean that it can be used to power other applications where the capacity factor has a meaningful impact on the end cost, such as hydrogen production.
It produces heat as well as electricity, which can be used for various industrial applications, or space heating.
Its big turbines can provide inertia to help balance the grid (although there are already alternatives being rolled out)
It has low material input (and therefore low lifecycle emissions), and doesn’t contribute to massively increasing demand for copper, for example.
It has the least land use - by far - of electricity generation, and certainly zero-carbon electricity generation (except rooftop or integrated PV, of course), at about 1/300th of solar PV. Nuclear the green dot below, note is log-scale:
How much does nuclear already contribute? A significant amount. Nuclear energy has been contributing between 2400 and 2700 TWH of electricity over the last decade (dipping in the middle and then rebounding). As a percentage of global electricity supply it represented almost 12% 10 years ago, and has fallen to just under 10% today. By contrast, generation from wind and solar rose from 500 TWH to 2800 TWH and its share of generation rose from under 2.5% to over 10% in the same period, thereby overtaking nuclear as an electricity source in 2021.
So over 10 years, nuclear production is flat while wind + solar production has more than quintupled? Does that mean the game is up for nuclear? I find this “pace of deployment” argument against nuclear perhaps the most compelling. Whilst nuclear advocates have for years understandably pointed to how little relatively wind and solar have contributed, that is a backward snapshot, with the inflection point now reached and expected to extend significantly. To quote Amory Lovins from his conversation with Michael Liebreich (covered here):
“In 2020 the world added 0.4 gigawatts more nuclear capacity than it retired, whilst the world added 278 gigawatts of renewables – that’s a 782-fold greater capacity. Renewables swelled supply and displaced carbon as much every 38 hours as nuclear did all year. Where nuclear is cheap, renewables are cheaper still and efficiency is cheaper than that. There is no new type or size or fuel cycle of reactor that will change this. Do the maths. It is game over.”
Lovins’ (extremely negative) views on the prospect for nuclear are elaborated in this article.
Even adjusting for nuclear’s higher capacity factor, that still means that renewables installations outstripped nuclear by x200. Is it game over for new nuclear then? It seems not.
Nuclear has clearly moved up the agenda for governments. The UK has just recommitted to building nuclear, aiming to approve 8 new reactors this decade and for installed nuclear capacity of 24GW by 2050, vs today’s total of 7GW. (Capacity factors for nuclear is about twice that of offshore wind in the UK, so equivalent to about 50GW of offshore wind generation capacity, which, as it happens, is the UK’s 2050 target for offshore wind…)
Meanwhile France, which already gets 70% of its electricity from 60GW of nuclear capacity, has said it will build another 6 and potentially 8 more behind that of the existing light water reactor design the EPR2. This is the same design that is being constructed at Flamanville, which has just announced yet another delay and whose costs have overrun original plans by x4 (!!). Like the UK, it has also said that it will push to develop SMRs.
And China, as per usual, blows everyone out of the water, with as many plants in various stages of development as the rest of the world combined (below figures don’t include recent updates from France and UK):
Current construction: There are currently 55 nuclear plants actually under construction globally, mostly in China and India, bringing about 55GW of capacity online over the next 5 years or so. There is a further 45 planned (assuming here meaning they’ve received final approval) to bring total new capacity up by 100GW.
However, even in China, where there is by far the largest build out, nuclear is being far outpaced by renewables construction.
What about SMRs? The promise of small modular reactors, is that their deployment can shift nuclear from idiosyncratic civil engineering infrastructure projects to repeatable, mass manufacturing proposition, making the components in factories and starting to benefit from the attendant learning curve. There are a number of companies working in the space, but none will be benefiting from a strong learning curve any time soon with the first demonstration reactors live by the end of the decade at best. SMRs were prominently featured in The Economist recently.
NuScale - the only SMR company to have been approved by the US’s Nuclear Regulatory Commission (NRC), recently went public via a SPAC merger. NuScale recently announced a project in Poland to build a plant up to almost 1GW (i.e. not very “small”) “as early as 2029” with a partner in the mining industry who notes that nuclear is attractive as it produces heat as well as electricity. They are also developing a project in Idaho, one in Kazakhstan and one in Romania, all 462MW, all set towards the back end of this decade. Nuscale is a modular version of the common light water reactor.
TerraPower - founded by Bill Gates, TerraPower has a molten chloride fast reactor design, which runs efficiently at higher temperature and leans on the capacity to provide high-grade process heat as well as thermal storage, adding additional flexibility to ramp output up and down. One of two winners of a grant from the DoE under their Advanced Reactor Demonstration Program (ARDP). They are working towards repowering a coal plant in Wyoming. [An excellent application for nuclear where the transmission infrastructure can be reused.]
X-Energy - the other recipient of the ARDP, X-Energy is also using high-temperature reactor, together with high-refined fuel (HALEU) and is working towards a first plant of up to 320MW in Washington state.
Oklo - making micro-reactors, recently had its application to the NRC denied for failing to provide enough information around key safety features. Planning to make micro reactors (think 10MW or 1/100th of the size of typical nukes today) using a fast breeder reactor that allows them to use waste from other reactors as fuel. The COO appeared on MCJ podcast a while back. (I have notes on that from pre-substack days if anyone is interested.)
ARC Clean Energy - Another fast breeder reactor, 100MW this time, Canadian company with a salt-coolant (I think like TerraPower, which also uses salt) aiming to repower retiring coal plants (also like TerraPower).
Ultra Safe Nuclear - like Oklo, a micro reactor design, with individual units 5-10MW, initially targeting remote, solar-poor communities (Alaska). They have submitted their intent to submit an application to build a 15MW test plant, but doesn’t look like application proper has started. Another cool thing about USNC - they are developing nuclear thermal propulsion engines for space. The Jetsons nuclear future is coming!
Rolls Royce SMR - not a start-up, of course, but received funding from the UK as part of a consortium to advance its design. On the big side for “small” at 470MW. The Economist article linked above quoted the company as saying it was the biggest they could make it, whilst still fitting all the components on a truck.
How about the fuel? How much uranium do we have? Current identified reserves at economic (<$130 / kg) stand at about 6 million tonnes, or nearly 100 years of supply for today’s 400GW nuclear fleet. There is a further 2 million tonnes or so identified at higher-cost of extraction. On this, it is worth noting that we’re pretty good at improving extraction methods for resources in general, allowing us to continually bring down the cost of extracting the less readily available resources, e.g. shale oil. We can make a lot of headway in extraction and fuel efficiency during the course of the next decades. Eventually, someone might crack extraction of uranium from sea water, which exists at extremely low concentrations (3 parts / billion), but effectively contains a limitless supply if we can work out how to get it. (It’s very early days but progress is being made.) Also, as noted above, some of the new reactor designs use fuel much more efficiently.