"The obvious solution — more solar farms and wind turbines — faces significant obstacles. "

Like economics and common sense. Current meteorological conditions are a case in point - an area of high pressure means there's stuff all wind output across all of Britain (and Ireland, and most of Europe), and high levels of prevailing cloud cover (and seasonal aspects) mean solar provides little by way of alternative in the UK. So as of my writing this, if we ignore the faux-renewable output of Drax, then barely 10% of total electricity demand is being met by renewables.

Perhaps the article author can cost up the building of (a) sufficient storage to cover periods of dunkelflaute, and (b) the additional wind power assets to power us through these sort of periods. What's needed would be sufficient storage to be able to despatch a peak demand of the order of 40-50GW (far more if heating is electrified), and to have the ability to cover a dunkelflaute period of up to say nine days.

https://gridwatch.co.uk/

Shirley they will just suck the juice out of all the plugged in electric vehicles when the grid falls a trifle short.

At least that's what they were telling us a few short months ago ...

"Shirley they will just suck the juice out of all the plugged in electric vehicles when the grid falls a trifle short."

That's definitely possible and the technology is proven, but when factoring in all the availability and efficiency aspects then even when there's a sizable EV fleet the vehicle to grid resource is a tiny fraction of the scale required by a largely renewable grid if wind output craters. As a guide, V2G might offer at most 3-7% of the storage capacity to power most of the grid.

@Like a badger

"Perhaps the article author can cost up the building of (a) sufficient storage to cover periods of dunkelflaute, and (b) the additional wind power assets to power us through these sort of periods."

Reading the article I am not sure if it should be taken as his honest opinion or with the expectation of dripping sarcasm. It certainly reads both ways, especially how we have replaced energy generation with these things that dont work very well.

This line made me laugh- "What's less clear is how to reach this renewable utopia given decades of infrastructure underinvestment" considering the vast investment in the sky god monuments in place of power generation. Another reason to think Dan is being very sarcastic.

Dan if you intended your article to be read with the tongue in cheek I would like to say it was a great read and rolling my eyes at the current state of things here.

A fair comment, and apologies to Dan if I've taken his words out of context. There is though the unfortunate reality that politicians (particularly most Secretaries of State for energy matters) do think that it's as simple as throwing yet more bill-payer's money at wind and solar.

That's because all politician's are unfit for the positions they are given. We don't have engineers in charge of energy, we have people who went to school to learn to talk and be politicians.We wouldn't let them wire our garden shed yet they're entrusted with guiding an entire population to an unknown future!

"Small modular reactors (SMRs) are gaining attention from government and datacenter operators, but the technology remains largely untested, "

This sort of statement appears all over the place when SMRs are discussed and it puzzles me. Doesn't every nuclear-powered ship and sub have a small reactor in it? They might not be modular, but they're small, reliable and the safety work has been done already. Is it just a matter of power - which I assume is fairly scaleable - or am I missing something more fundamental?

You're right but it's not as simple as that. Lots of things on warships don't have to meet civilian safety standards, and access to cooling water is less of a critical issue for a submarine. Before they could be deployed across the countryside, either SMRs or nuclear safety regulations would need significant redesign.

Cooling isn't much of an issue: Just build either at existing coastal big nuclear sites, or at former coal power stations which all used river water cooling (hence the huge cooling towers). In both cases there's existing grid export links as well.

I think the problem is that although SMR in military service are well proven and easily joined together, their cost relative to output isn't good, and will be made dramatically worse by the appalling slow, interventionist and incompetent civil nuclear regulation.

As an example there's the dispute over around 7,000 design changes at Hinkley C, that Areva claim resulted in 35% more steel and 25% more concrete than the reference EPR design. The Office of Nuclear Regulation disputes this, but SOMETHING has caused EDF to spend a shed-load more on changes. Since EDF are on a fixed output price contract, we can safely assume they didn't do it of their own volition.

The SMRs being proposed are mostly light water pressurised reactors. Not exactly revolutionary. The difference between them and a sub reactor is that the SMR would be run on low enriched fuel, the same as most currently operating commercial reactors. The sub/ship reactors are usually run on high enriched uranium which is VERY expensive. The 'untested' bit will be the supporting infrastructure. Containment building, safety systems, cooling, refuelling etc.

These 'small' modular reactors are not actually that small. The RR SMR is proposed at nearly 500MWe. Calder Hall was 60MWe per reactor and the first GE BWR was 160MWe per reactor.

Obsolete nuclear submarines tend to be quietly scuttled in deep water. Nobody decontaminates old reactor compartments

I disagree with the bullshit AI bollocks. I do find it useful however, to help, not fully write, but help with doing scripts. You, of course, need to understand what the code is doing as sometimes its wrong and dangerous but it helps. But its better than being told by the likes or reddit or stackoverflow "If you don't know what you're doing, you shouldn't be touching it".

Anyone, so, despite mostly disliking all the AI bollocks, Wales would be a good place for them to start. Wales need/want the jobs and they've got lots of renewable energy in water.

"Wales need/want the jobs and they've got lots of renewable energy in water."

Two problems, (1) there's not many long term jobs in any form of renewable power or datacentres, and (2) Wales doesn't have any "spare" renewable power.

On point 2, according to the Welsh government, Welsh electricity demand is about 15 TWh, and Welsh renewable output from all renewable sources is around 8 TWh.

Right now there is a lot of discontent over plans to build massive pylons to bring power from offshore wind across the landscape of South West Wales. Given that the plan is presumably to burn all of that electricity in datacentres in England, building the datacentres here and pushing the data across the country instead is an unbelievably obvious solution to a whole bunch of related problems.

Well there's a lot of roof space, why not place solar panels on those, granted not all roofs are suitable but its a lot of land unused. Also, lets not forget that ghoul Thatcher created most of this when she sold our infrastructure even though she's dead her policies still haunt our economy from beyond the grave, and pre-pandemic the Tory government was selling off the gas storage which we rely upon, so when the gas prices rise it hits us hard.

In gold rush times the most money was made by the spade manufacturer, now in this instance what is the equivalence of the spade?

"Well there's a lot of roof space, why not place solar panels on those"

We could, but as its small scale installations the unit cost is fairly high and the power resource is poor. An expensive asset like a data centre wants to run at north of 75% utilisation, the equivalent metric across the year for solar is capacity factor and that's about 10% in the UK, much lower in winter, and obviously zero output when the sun has set. Could offset a proportion of gas powered peaking, but on its own that's not going to make an investment case.

Now, if we were in Nevada where peak demand is in summer and heavily influenced by aircon, then (a) solar capacity factor will be around 30%, and (b) output will be very strongly correlated with peak demand. So in the right place solar is a no-brainer. But here in the UK not only is the output puny, it is inversely correlated with both day time and seasonal peaks.

"In gold rush times the most money was made by the spade manufacturer, now in this instance what is the equivalence of the spade?"

GPUs, and I think to a lesser extent high performance memory. In 2011-2012, Nvidia sold $4bn of product and had a net profit of around half a billion. In 2024-2025 they sold $130bn of product and had a net profit of $73bn.

We are going to be ready for the AI gold rush just about a week after the bubble bursts and the whole thing collapsed aren't we..

…need to be in the UK. The computers can be anywhere in the world.

Can I recommend somewhere with lots of sun? Maybe the Sahara desert.

Stupid politicians….

There are schemes mooted to build huge solar farms in the Sahara, and use LV DC to transfer the power to Europe. Problem is that being dependent on the unstable and undemocratic countries of the Sahara whether for power (or compute power) adds even more geopolitical challenges than exist with Middle East oil and gas. Moving compute workloads that distance creates latency challenges and data sovereignty issues in countries that rate poorly for transparency and corruption.

At large scale there's also the fact that really large scale solar panels would have a climate heating action due to their lower albedo than sand. Probably still better than fossil fuels, but it does close the gap.

While the bullshit-generation is exciting and impressive, it's of very limited use. But the hype has forced politicians to think seriously about how the UK needs to power itself in the next few decades: a Very Good Thing because otherwise these long-term issues are mostly ignored by 5-years-max Parliaments.