Or just batteries? Throw on a gas turbine emergency reserve running your favorite fossil or green fuel for well, the emergencies. We’re talking irrelevant emissions.
I truly can’t comprehend where this massive boner for new built nuclear power comes from. Sci-fi?
ah yes, green fuel like dirt cheap H2 and irrelevant emissions from multi GW of gas/coal firming...
I'm not sure what you mean by a boner for nuclear, but we see France has one of the lowest emissions in EU, did teh job in under 20y, is largest net exporter in EU, and spent on this half of what DE spent on EEG alone. Germany is still one of the worst emitters in EU after 25y of ewende.
No country managed to match french emissions (without sufficient hydro firming) even now. Denmark(albeit helped by nordics), California and SA still have higher emissions
More like a few decades, if ever. Battery production is estimated to reach 6.8 TWh per year in 2035 [1]. But only 12% of this is expected to go to battery storage systems, yielding just 800 GWh. 12 hours of storage worldwide at current electricity demand is 30,000 GWh. And by 2035, electricity demand is going to be more than that, as transportation and industry is increasingly electrified and migrated off of fossil fuels.
To reiterate, this is just for 12 hours of storage. Seasonal fluctuations can depress renewable generation for days or even weeks.
wind runs through the night, and so does hydro and existing nuclear. So lets say, 15,000gwh that needs shifting? Still a big number, but...
BESS increased 45% y/y in 2025, and is looking like higher growth than that in 2026 already ~60% (1). Im optimistic that the mckinsey conservative linear estimates of growth are missing the s-curve of scaling new tech, just like they did for solar. They only have to be wrong by a little y/y and we get to 1000ghw a year by 2030 (note, they released a more recent study that pulled 800gwh/year in to closer to 2030 [2], the previous study was already too conservative) . At 1twh a year we're seriously chipping away at emissions, we're done in 15 years if nothing else changes (which of course it will, both on the demand and supply side). Still, thats actually incredible!
Wind doesn't cut out at night, but it also experiences long periods of low production: https://www.eia.gov/todayinenergy/detail.php?id=46617 It is unquestionably an intermittent source that would require overprovisioning and large amounts of storage to even out periods of underproduction.
The projections for battery growth might be off, sure. But it's also possible the growth is a little bit under the projections year over year, and then we're looking at much less battery production five years. You're invoking uncertainty, but only considering it in one direction.
Large geographic networks like the EU really help to smooth that variance out. Anyway, on average is all that really matters here. Remember, I’m not saying we can only have batteries and if they run out we’re in the dark. We’ll use gas to fill the gaps, and emit emissions for it. All that matters is the total emissions for the year.
Yes, im explicitly considering it only in one direction, as I said I’m optimistic. I have seen plenty of data, my own & others reasoning that leads me to believe in the optimistic case here.
EDIT: look at the graph in that second McKinsey link. Look at the step for 2024, and then the massive step for 2025. And then they project much tinier steps for 26 and beyond? That’s obviously nonsense. And we can tell it’s nonsense as the number for 26 are coming in at another 60% increase y/y, and all reports point to huge increases in deployed capacity this year. And they have it at like 20%. Cmon, that’s nonsense.
That link is a graph of battery electric storage, specifically. Mckinsey's projections have battery production continuing to accelerate, but the lion's share of the output dedicated to electric vehicles rather than grid storage: https://www.mckinsey.com/features/mckinsey-center-for-future...
Rather than just proclaiming the projection as "nonsense" it'd be a lot more productive if you shared an alternate projection and explained why it's methods are superior.
We of course can’t scale the grid portion of battery production as fast, or even faster than what we’ve done for BEVs?
And this also disregards that second life automotive batteries are incredibly hot on the market. All those TWh of batteries will become available for stationary use when the cars are scrapped.
Maybe not in western markets due to labor costs, but definitely in developing economies.
Used automotive batteries will be at the end of their life, with only a few hundred more discharge cycles until they've totally degraded.
Again, batteries scaling rapidly still doesn't hold match up with the scale of electricity demand. Again, a day's worth of global electricity consumption is 60,000 GWh. If there's one lesson to take away, it's this: be skeptical of people who talk about "scale" but neglect to actually give figures for that scale.
All recent research coming are showing that BEV batteries last longer than expected in real world conditions.
You do realize that with 60 TWh we’re arguing about which decimal of 99.x% renewables the grid sits at?
We have in a few years scaled BEVs to 3 TWh per year. For grid duty batteries last 10-15 years. They are essentially the same batteries. Some have different form factors and whatever, but the core is the same.
This seems like grasping for the straws. Denying what’s infront of your eyes because we still need a few more years of scaling until it will happen no matter what, just assuming a continued buildout to saturation.
The grid works on timescales of decades. With the current deployment rate, no matter how you try to belittle it, we’ve already locked in a complete revamping of our grids.
Global battery deliveries in 2025 was 1,600 GWh, not 3,000 GWh. Of that, only 300 GWh was used for grid storage. Even projections a decade out to 2035 only predict that yearly battery storage deployment will be 800 GWh per year, the vast majority of battery production will be used for EVs rather than grid storage.
Again, there's a reason why people talking about batteries scaling don't put their numbers in the context of electricity demand. Even with the predicted exponential growth of battery production, the scale of electricity demand is on a different level.
Or an iron air. Or flow battery. Or sodium. Or all manner of different lithium chemistries across NCA, NMC, LFP and so on.
We’re seeing the Cambrian explosion of battery technology. From early BEVs utilizing the highest performance to even deliver a viable product to a plethora of options depending on your use case.
How about we don't bother with either fission or wind and solar? Just build fusion plants and be done with it. If we're comfortable betting trillions of dollars in infrastructure projects on the hopes that a heretofore unproven technology will pan out, let's be more ambitious than batteries!
Of course, there's no guarantee that any of those fission ideas will actually pan out. Likewise with these battery chemistries. Investing loads of money into intermittent sources with just the hope that some future battery chemistry will solve storage at grid scale is not what I'd call a wise plan.
Anybody who over the past few decades has been saying that we could not deploy batteries on a massive scale needs to reevaluate their bad assumptions, because they are wrong, and moreover we should not trust any of their current assessments until they can reconcile what they got wrong. The tech curves of batteries have been clear for decades, this tech development should not have been unexpected.
Your link reports that the USA added 15 GW of battery storage in 2025. I'm not sure how this is supposed to demonstrate the feasibility of battery storage at grid scale. Let's actually express the scale in terms of numbers relative to our electricity demand:
* The USA uses 12,000 GWh of electricity per day
* The world uses 60,000 GWh of electricity per day.
* Global battery production in 2025 was ~1,600 GWh, of which 300 GWh was used for grid storage [1].
At our present production rates, it'll take 100 years to provision 12 hours worth of storage at 300 GWh per year. Batter production is set to increase to 6.8 TWh per year [2], but only 12% of that is predicted to go to grid storage, or about 800 GWh per year. Even at 2035 rates, we're looking at 37 years of production to fill 12 hours of storage (12 hour of electricity storage for 2025 electricity demand rates, which will be higher in 2035).
Yes, batteries are being deployed at a massive scale today. But electricity generation is on an even more massive scale that dwarfs battery production rates.
> Your link reports that the USA added 15 GW of battery storage in 2025. I'm not sure how this is supposed to demonstrate the feasibility of battery storage at grid scale.
Mmmhmm, grid scale deployment is not grid scale now? You are redefining terms, which means you don't work in the field and are not at all familiar with the field, yet you make broad and sweep proclamations of incredulity that have no factual backing, and we are supposed to trust you purely on judgement?
You cite last year's deployment rate, without noting a massive increase in planned deployments for this year. You neglect to cite the year before it, which was much smaller. You looking at a puck headed to the goal, under a continuous accelerant force, and saying, "the puck is here, therefore it will never hit the goal." That's a ridiculous thing to assert, because you don't hold that afactual standard for any other technology, just batteries, yet seem to understand that all other technologies have continually changing amounts of producition.
BTW, your link is "demand" and disagrees with most other sources.
> At our present production rates
That kind of says it all, doesn't it? You think that present production rates are indicative of future production rates, which is an insane statement.
If nuclear has 0 GW new this year, how do you think it could ever get to 2GW/year, right?
You have no reasons for these strange beliefs that defy data and trends, you just assert incredulity. It's completely irrational.
Again, you have to put the scale of battery production in the context of electricity demand. 300 GWh of battery storage being provisioned sounds like a lot until you put it in the context of 60,000 GWh of electricity consumed daily. There's a reason why proponents of battery grid storage never actually put their numbers in the context of electricity consumption.
I'm not expecting readers to trust me purely on judgement, I'm expecting them to do the math and realize that battery storage deployment and electricity demand are multiple orders of magnitude off, even with the projected increases in battery projection.
> That kind of says it all, doesn't it? You think that present production rates are indicative of future production rates, which is an insane statement.
Again, I did cite the projected production figures for 2035. Did you miss that part?
We've also performed fusion in a lab. That doesn't mean it'll be viable in production at scale.
What's the annual production figures for iron air batteries, flow batteries, etc.? Sodium batteries are at 9 GWh delivered in 2025. Google tells me that flow battery capacity is 500 MWh to 1 GWh, but doesn't provide any figures on actual production deliveries (production capacity is not the same as actual delivered production). There are no iron air battery facilities currently in production, with the earliest plant trying to open in 2028 with 500 MWh per year annual production.
None of your suggestions are remotely close to operating at grid scale, and there is zero guarantee that any of them will prove more feasible than lithium based battery chemistry.
This sort of moving the goal posts is not convincing at all. First it was "batteries will never scale to grid usage" now it's "early days of production of a brand new chemistry are only at 9GWh". You seem to think that is somehow an indictment of the technology rather than a statement of an amazingly quick scale up, with no signs of stopping. That's just bad judgement to say "a rapidly scaling tech is at GWh scale even without much demand therefore it's useless".
Meanwhile, the statement that "fusion has been achieved in a lab" is optimism and wishful thinking beyond words. What energy return did that get? What was the cost? When will there be GE of generation, mic less basic safety engineering?
Those who advocate against the shipping reality of batteries, and moreover assume that they will get more expensive, are not using numerical thinking and are not thinking like scientists, engineers, or technologists. They are merely rooting for a tech like a sports team. Nuclear does not need any more fans, it needs engineers and scientists that can achieve some sort of radical breakthrough that makes it a desirable tech.
> First it was "batteries will never scale to grid usage"
You're inventing a straw man that's easier for you to attack.
No goalposts are being moved. My point was, and still is, that batteries do not presently scale sufficiently to make a predominantly wind and solar grid feasible, and our current projections even a decade out do not see them scale to that point either.
We don't know if some breakthrough in battery chemistry will make it scale. Could such a breakthrough transpire? Sure. But will it happen? We don't know. And thus we should not gamble massive infrastructure spending on the assumption that this breakthrough will happen.
Even with Danish insolation and weather and tilting the study heavily towards nuclear power by assuming that the nuclear costs are 40% lower than Flamanville 3 and 70% lower than Hinkley Point C while modeling solar as 20% more expensive renewables come out to vastly cheaper when doing system analyses.
This paper is very heavily biased against nuclear power and is only valid for Denmark
It uses 8% discount rate for nuclear vs 5% for VRE
It uses the most expensive nuclear reactor costs instead of Korean and Chinese reactors delivered at 3,500–5,000 USD/kW
80% capacity factor for nuclear is very low and should be over 90% for new reactors.
It's least cost mix intentionally excludes nuclear power which is absurd. Standard practice would let the optimizer choose nuclear's share in a hybrid mix. Sepulveda et al. (MIT, Joule 2018; Nature Energy 2021) using exactly this approach repeatedly find firm low-carbon resources (including nuclear) reduce total system cost under deep decarbonization. https://www.eavor.com/wp-content/uploads/2021/11/The-role-of...
"Availability of firm low-carbon
resources reduces costs 10%–62%
in zero-CO2 cases"
They intentionally ignore inter-annual variability which is where dispatchable nuclear is most needed.
It generalizes based on Denmark's unique situation of having some of the best off-shore wind in the world and access to cheap hydro power and storage in Norway and no domestic nuclear supply chain.
The authors are editors of the journal this was published in.
Lund is the creator of EnergyPLAN and cites himself a lot.
This paper just repeats Aalborg group and Breyer's LUT group's anti-nuclear opposition.
Hinkley Point C and EDF just got a 7% interest rate to finish the project. That is after nearly 20 years of project work and 10 years of construction, so about all risk should already have been found.
Like I said. The costs are 40% lower than Flamanville 3 and 70% lower than Hinkley Point C.
Imaginary cheap and fast to build nuclear power is amazing. It also does not exist. In South Korea those costs are from before the corruption scandal.
In China they are barely building nuclear power. It peaked at 4.7% of their grid mix in 2021 and is now down to 4.3%. For every plan they release the nuclear portion shrinks and is pushed further into the future.
Then I just see you trying to handwave the study away. The entire point is literally to prove that Denmark does not need to rely on its neighbors, and still get a cheaper result.
And like I said. Denmark is the hard case due to the winter sun being awful. As soon as you go south in latitude the problem becomes vastly easier. We’re talking like 99% of the worlds population having more sunlight than Denmark.
"Imaginary cheap and fast to build nuclear power is amazing."
It isn't imaginary. Korea and China prove it is possible to build nuclear reactors for reasonable cost when you don't have endless irrational legal opposition that makes them take much longer to build. What IS imaginary is multi-day grid scale storage. All BES are designed with at most 4 hour capacity.
I didn't handwaved away the study I carefully pointed out how it is systematically biased against nuclear which isn't surprising considering how anti-nuclear the authors are.
Denmark isn't nearly as hard of a case as you think because it has some of the most reliable off shore wind power available.
And it's conclusion about Denmark, if correct, cannot be generalized to the rest of the world. You have to have dispatchable power in an electrical grid and that has to come from gas, coal, or nuclear.
Even with Danish insulation and weather and tilting the study heavily towards nuclear power by assuming that the nuclear costs are 40% lower than Flamanville 3 and 70% lower than Hinkley Point C while modeling solar as 20% more expensive renewables come out to vastly cheaper when doing system analyses.
This article conveniently doesn't include flexible demands and energy storage, both of which are a solved problem with nuclear but completely unrealistic with renewables.
Nuclear does neither flexible demand nor energy storage, those are in fact the things that nuclear does not solve! There are a few flexible nuclear plants in France but they push up costs. Some of the new modular nuclear rector designs are considering storage/flexibility, but cost there is also expected to be far higher than an AP1000.
Batteries are cheap, getting cheaper, and are the biggest disruption and innovation on the grid in more than half a century. You can use them to reduce transmission costs, to reduce the load on distribution substations and increase distribution usage capacity, you can use storage to make everything a lot cheaper by allowing smaller sizes for expensive T&D equipment that sees less than 30% average utilization.
Calling batteries "unrealistic" is not based in reality, it's just being stuck in decades past.
Nuclear can deliver flexible demand if required, it just involves either lowering the reactivity in the core, or if the drop in demand is sudden, bypassing steam from the turbine and running it directly to the condenser. But since their operating costs are so much lower than their construction costs operators run them at 100% capacity as long as they can.
The only situation where deliberately operating a nuclear plant at under 100% output is when nuclear makes up a very large chunk of a country's generation capacity. It's not that only French nuclear plants can reduce output it's that only the French have ever been in the situation where their nuclear capacity exceeds their minimum electricity demand.
Not economically. EDF is already crying about renewables cratering the earning potential and increasing maintenance costs for the existing french nuclear fleet. Let alone the horrifyingly expensive new builds.
And that is France which has been actively shielding its inflexible aging nuclear fleet from renewable competition, and it still leaks in on pure economics.
It's never economical to operate an asset at under 100% capacity. Intermittent sources of energy like solar and wind encounter the same problem when they start to saturate demand during peak periods of generation. Install a new solar panel in California, and chances are you won't actually be able to sell any electricity around midday since demand is saturated.
The difference is that nuclear will keep running at night, in the winter, regardless of how strong the wind is blowing. A cheap, but intermittent source of carbon-free energy is not comparable on a dollar-by-dollar basis to a non-intermittent source of carbon-free energy.
The common retort is to use batteries, but let's put this in perspective: France uses 1,219 GWh of electricity daily (note that this is just electricity and doesn't include things like transportation, fuels in smelters, chemical feedstock etc.). 12 hours of storage would be 600 GWh. Seasonal fluctuations in wind and solar are even more extreme, and might need days worth of stored energy. But let's be humble and just see what it'll take to provision 12 hours:
At $150/kWh that'll be 90 billion dollars. These batteries will be good for 2,000 to 5,000 cycles. Let's say 4,000, so it has an 11 year life span. Over the course of 55 years that'd cost $450 billion. Just for the storage, mind you, France has to build the renewable generation on top of the storage.
On the flip side, the Flamanville Nuclear plant has a lifespan of 60 years. You could build 12 Flamanville nuclear plants and satisfy 100% of France's electricity demand. At €19 billion euros, or about $22 billion USD building 12 Flamanville plants would work out to $264 billion. The cost of storage to even out intermittent sources is much more expensive than just building the nuclear plants.
That ignores operating costs and battery costs are falling fast and your assumptions seem overly pessimistic. A 2025 project in Italy came out at $120/kWh made up of $70/kWh in equipment and $50/kWh in engineering and grid connection costs. (The grid connection will still be good and concrete pads can be reused so replacing after 20 years will cost less even before price drops in equipment.)
Even with a 7% cost of capital that gives a levelized cost of storage of $65/MWh or an additional $33/MWh on top of the levelized cost of electricity of solar to spread it across day and night [1].
With a 4% cost of capital the still being designed EPR2 with 30% savings over Flamanville 3 comes in at €93/MWh or $110/MWh [2].
So solar costing less than $77/MWh or €66/MWh + storage should be cheaper than EPR2.
The same applies to nuclear power, though: when France built multiple copies of the same plant design, the first few builds were expensive but costs declined for subsequent models. It's fine to include projected costs reductions into your cost estimate, but you have to apply the same logic to competing systems.
These numbers already include a projected 30% drop in costs for EPR2 across building six reactors with the first coming online in 2038.
Building a series of nuclear reactors with overlapping schedules (about one completion every year or two) in one country should help. But it’s simply far easier to find cost reductions for wind turbines which are manufactured in the thousands per year or solar panels and batteries which are manufactured in the millions.
Or you know, just build renewables and storage. Displace vastly more coal faster with a death per kWh where the only injuries comes from traditional construction and mechanical industry work.
evacuation in Fukushima was forced. They could have stayed just as well - dose was too small. Statistically nuclear is great. Even more so considering fossil firming strategy in many countries
That is not the case. Grid modelers always land on renewables being cheaper. Except for the cases when the studies start with "assuming cheap and fast to build nuclear power".
That is not the case. There are LCOLC, LFSCOE and others which land on renewables being way more expensive. Even without your made up claim about "assuming cheap and fast to build nuclear power".
Like the LFSCOE study is only using one source of renewables through all weather together with 2020 data on battery costs.
Which is why I linked a recent full system analysis. With Danish data so a vastly harder problem than a place with abundant solar. So tell me what they missed.
They even tilted the study heavily towards nuclear power and assumed that the nuclear costs are 40% lower than Flamanville 3 and 70% lower than Hinkley Point C while modeling solar as 20% more expensive.
Still finding that renewables are vastly cheaper when it comes to meeting a real grid load.
you should be very careful with 'papers' written by people that stayed at the core of Danish antinuclear movement like Henrik Lund (famous “for me, nuclear power is not a dream. It’s a nightmare. My dream is renewable energy”, but there are more interesting turns there, including EnergyPLAN which was criticized a lot for it's nonsense assumptions which I can bet is used here too. It was so severe that the issue was addressed in a peer reviewed publication https://www.sciencedirect.com/science/article/pii/S030626192... ) and a paper that includes frauds like jacobson in citations. These two are famous for citation rings and both are public antinuclear activists.
I can bet they have very "optimistic" estimations for Hydrogen and gas firming on top of the most evident issues adressed in peer review
At a cost which could generate ~10 billion watts of very low CO2 electricity for decades if invested in renewables.
Also remember that large parts of a nuclear plant is replaced over its operational life. Control systems, steam generators, turbines, generators, tubing, valves etc.
What stays is the outer shell and pressure vessel. A nuclear plant doesn't just "work" for 60 years. And there's no trouble designing renewables with a 60 year lifespan.
We just don't do it because spending money on getting their expected operational lifetimes from decades to 60+ years is betting on extremely uncertain future returns.
Did we have rolling blackouts from electricity shortages during the energy crisis? No.
Was the electricity extremely expensive? Yes.
Reliable electricity has a certain worth. And that is vastly lower than what nuclear power needs when running at 100% 24/7 all year around.
And that is disregarding that EDF is already crying about renewables crater the earning potential of their existing nuclear fleet due to load following and increased maintenance costs. Let alone horrifyingly expensive new builds.
but for ren you need parallel gas firming. For nuclear you need some backup, but not fully parallel grid. Paid off npp can generate very cheaply, at 4-7ct/kwh
I don’t see the difference with nuclear power? Take California, a yearly baseload of 15 GW and peak load of 52 GW. What problem is even a ”baseload” of nuclear power solving?
But we should of course keep our existing fleet around as long as it is safe, needed and economical. In that order.
EDF is already crying about renewables cratering the earning potential and increasing maintenance costs for the existing french nuclear fleet. Let alone the horrifyingly expensive new builds.
And that is France which has been actively shielding its inflexible aging nuclear fleet from renewable competition, and it still leaks in on pure economics.
France does not have a fully parallel nuclear power grid ready to step in when some other half isnt working. Germany has a fully parallel fossils firming grid on top of their ren deployments.
EDF isnt crying. It's just treated poorly even by looking at ARENH tax which was replaced with another one this year, while ren business gets CFD's and curtailment payments.
French nuclear fleet is extremely flexible, RTE data is public. In fact, due to ARENH law EDF was forced to subsidize competition because otherwise that competition would not exist.
Where does this "need for baseload" energy come from? Baseload is a demand side concern. It can be fulfilled by any number of sources and we already have grids operating with zero baseload.
The grids have dispatchable power. But that is a different concerns.
You also have to look at it in terms of outcomes. How do we get the most decarbonization the quickest per dollar spent?
Focusing on reducing the area under the curve. Looking at it from that perspective wasting money and opportunity cost on new built nuclear power leads to spending longer time entirely dependent on fossil fuels.
After SL-1 we realized that that we needed to allow a reactor to fully shut down even with the most important control rod stuck in a fully withdrawn position.
I used to work at Boeing on airliner design. The guiding principle is "what happens when X fails" and design for that. It is not "design so X cannot fail", as we do not know how to design things that cannot fail. For Fukushima, it is "what happens if the seawall fails", not "the seawall cannot fail".
Airliners are safe not because critical parts cannot fail, but because there is a backup plan for every critical part.
Venting explosive gas into the building seems like a complete failure to do a proper failure analysis.
>at Boeing on airliner design. The guiding principle is "what happens when X fails"...Airliners are safe not because critical parts cannot fail, but because there is a backup plan for every critical part.
And yet creating a culture that is vigilant and consistently applies due diligence is hard. To that point: Boeing identified the 737-Max MCAS as 'hazardous' in their analysis. Putting aside that 'catastrophic' was the more appropriate rating, they still did not appropriately design their system when that system failed. (By their own processes, 'hazardous' meant it should not be designed with single-point hardware failures)* That implies it is as much a human/cultural issue as a technical one.
* before any claims that the system was designed just fine because the pilots could have avoided the issue with the appropriate actions, those are administrative hazard mitigations which are generally considered less desirable than hardware fixes, especially when engineering mitigations are already installed but not used. Removing the hazard >> engineering controls >> administrative controls >> PPE. To the GPP point, hindsight is easy, managing risk, people, and processes is hard.
Check the previous note I left above with the * on why that is considered a poor mitigation.
Administrative procedures are bad mitigations in general but especially bad when a) it’s a safety critical issue and b) the hardware for an engineering mitigation is already installed. That’s like saying death could have been avoided if people would have just packed parachutes (PPE). Maybe true, but bad hazard mitigation.
I truly can’t comprehend where this massive boner for new built nuclear power comes from. Sci-fi?
reply