Another thing to say - and I experience this in my own work - the amount of data centre development currently in the pipeline is substantial.
It’s akin to a property developer gold rush. Maybe it’s just the end of the 18.6 year land speculation cycle, maybe it’s real.
But we’re talking minimum 100MW per site. And if there’s no power available then they’re considering building out prime generation (eg gas)
Whether these two could actually benefit the grid is another question. Traditional DC might work to tier 3 (99.98%) or 4 (99.995%) availability. AI DC might be more flexible in that regard. Not all AI, but some flexibility exists because outputs are not instantaneous an SC batch processing and training can go on hold. So my question would be “can these sites, once functioning, be curtailed to support grid stability?” It’s not ideal. And it’s not how I would have designed a grid. I would have installed far more firm power instead of wind and solar.
Now the next obvious question. “How much does it cost?” If you’ve invested billions in your AI DC GPUs, you don’t want them sitting around idle, unless someone pays you. Or unless you agreed to sign up to terms which meant the power was curtailable. So it comes down once more to NESO, to the DNO & TNO commercial contracts when offering DC sites with “firm power” and the repercussions of withdrawing.
Let’s say we have 50 AI DC sites of 100MW each, and they can wind down their GPUs, cooling systems etc when they get the NESO call. And they’re big enough to get that call. Let’s say they can reduce by 40%. That’s 4GW.
So lots still in play….and should keep us busy for years 😁
My son works for a company that makes parts for Siemens gas turbines. Siemens' order book is full. That tells you all you need to know about what DC owners are planning to do.
Certainly there are people in the UK with functioning vision. Why are they not involved in the process? Is it because "the powers that be" don't like what those with vision see?
If the next winter is like December 2010 when snow blanketed many regions, and temperatures were >-10c on several days, given that EVs have grown in numbers, and wind and solar power has multiplied, would there be widespread blackouts? It seems that the supply of power is being run on 'a wing and a prayer'.
The seventh carbon budget seriously underestimates the cost of offshore wind power.
The budget assumes £1,860 per MW, declining to £1,120 per MW by 2050, compared to real capital costs that exceed £3,000 per MW and are rising as wind farms have to be constructed in deeper water and further from shore. The incorrect assumptions distort the analysis of costs and identification of the lowest cost option.
Onshore wind and solar are similarly underestimated.
Apart from not working, which is, of course, the most serious deficiency, the costs are grossly underestimated.
You are right. Far too many people misunderstand what the Boston Consulting Group study on cost curves was saying. It talked about the price of standardised components built in the millions becoming cheaper. It didn’t talked about full integrated bespoke engineered systems becoming cheaper. Renewable energy fall into that bespoke engineered system category - as every site is different and as you rightly point out the easy sites have already been built. “Learning by doing” and “cost curves” have become lazy props for think tanks and poorly informed economists who fail to establish what element of the cost is really going to fall to deliver on the claims that they wish to make.
I'm wondering why the discussion of hydrogen storage for backup. Isn't nat gas cheaper than hydrogen, even if it has to be imported as LNG? - just asking...
Also, how can hydrogen stored underground be relied upon? H2 is the lightest element, and will probably leak out of any underground storage. It can even leak out of pipelines, which is why we don't have H2 pipelines and why it's so problematic to transport it.
The only way I see this circle being squared is the euphemistically-named Demand Management. Claims of Britain being an ‘AI superpower’ or re-industrialisation are fantasy if the current plan is the energy system which underpins it. Only being able to do things at scale when the wind permits is pre-industrial and will result in a pre-industrial society.
Interesting Richard. There are plans afoot by NG to introduce more interconnectors. LionLink (Holland) and Nautilus (Belgium). A major part of the business case of these floor and cap proposals - will be resilience in my opinion, and if I gad the job that’s what I’d be focussing upon.
To build the OBC to counter Ofgems modelling, I suspect a big part of the “benefit to the billpayer” will be the lights not going out. Also the ability to export abundant North Sea wind power rather than pay curtailment, although this requires the current grid upgrades to have been installed.
Another thing to say - and I experience this in my own work - the amount of data centre development currently in the pipeline is substantial.
It’s akin to a property developer gold rush. Maybe it’s just the end of the 18.6 year land speculation cycle, maybe it’s real.
But we’re talking minimum 100MW per site. And if there’s no power available then they’re considering building out prime generation (eg gas)
Whether these two could actually benefit the grid is another question. Traditional DC might work to tier 3 (99.98%) or 4 (99.995%) availability. AI DC might be more flexible in that regard. Not all AI, but some flexibility exists because outputs are not instantaneous an SC batch processing and training can go on hold. So my question would be “can these sites, once functioning, be curtailed to support grid stability?” It’s not ideal. And it’s not how I would have designed a grid. I would have installed far more firm power instead of wind and solar.
Now the next obvious question. “How much does it cost?” If you’ve invested billions in your AI DC GPUs, you don’t want them sitting around idle, unless someone pays you. Or unless you agreed to sign up to terms which meant the power was curtailable. So it comes down once more to NESO, to the DNO & TNO commercial contracts when offering DC sites with “firm power” and the repercussions of withdrawing.
Let’s say we have 50 AI DC sites of 100MW each, and they can wind down their GPUs, cooling systems etc when they get the NESO call. And they’re big enough to get that call. Let’s say they can reduce by 40%. That’s 4GW.
So lots still in play….and should keep us busy for years 😁
My son works for a company that makes parts for Siemens gas turbines. Siemens' order book is full. That tells you all you need to know about what DC owners are planning to do.
Who in their right mind will pay OpenAI for AI tokens if they can only use Chat GPT when it is windy or sunny?
Certainly there are people in the UK with functioning vision. Why are they not involved in the process? Is it because "the powers that be" don't like what those with vision see?
If the next winter is like December 2010 when snow blanketed many regions, and temperatures were >-10c on several days, given that EVs have grown in numbers, and wind and solar power has multiplied, would there be widespread blackouts? It seems that the supply of power is being run on 'a wing and a prayer'.
The seventh carbon budget seriously underestimates the cost of offshore wind power.
The budget assumes £1,860 per MW, declining to £1,120 per MW by 2050, compared to real capital costs that exceed £3,000 per MW and are rising as wind farms have to be constructed in deeper water and further from shore. The incorrect assumptions distort the analysis of costs and identification of the lowest cost option.
Onshore wind and solar are similarly underestimated.
Apart from not working, which is, of course, the most serious deficiency, the costs are grossly underestimated.
You are right. Far too many people misunderstand what the Boston Consulting Group study on cost curves was saying. It talked about the price of standardised components built in the millions becoming cheaper. It didn’t talked about full integrated bespoke engineered systems becoming cheaper. Renewable energy fall into that bespoke engineered system category - as every site is different and as you rightly point out the easy sites have already been built. “Learning by doing” and “cost curves” have become lazy props for think tanks and poorly informed economists who fail to establish what element of the cost is really going to fall to deliver on the claims that they wish to make.
https://ukenergythoughts.substack.com/p/the-experience-curve-and-renewable?r=4ie0c9&utm_medium=ios
I'm wondering why the discussion of hydrogen storage for backup. Isn't nat gas cheaper than hydrogen, even if it has to be imported as LNG? - just asking...
Also, how can hydrogen stored underground be relied upon? H2 is the lightest element, and will probably leak out of any underground storage. It can even leak out of pipelines, which is why we don't have H2 pipelines and why it's so problematic to transport it.
The only way I see this circle being squared is the euphemistically-named Demand Management. Claims of Britain being an ‘AI superpower’ or re-industrialisation are fantasy if the current plan is the energy system which underpins it. Only being able to do things at scale when the wind permits is pre-industrial and will result in a pre-industrial society.
Interesting Richard. There are plans afoot by NG to introduce more interconnectors. LionLink (Holland) and Nautilus (Belgium). A major part of the business case of these floor and cap proposals - will be resilience in my opinion, and if I gad the job that’s what I’d be focussing upon.
To build the OBC to counter Ofgems modelling, I suspect a big part of the “benefit to the billpayer” will be the lights not going out. Also the ability to export abundant North Sea wind power rather than pay curtailment, although this requires the current grid upgrades to have been installed.
The issue with wind is that on the whole if the UK has a surplus, so does continental Europe, vice versa for wind drought.
Clearly that's not always true, but it's no way to build a resilient grid for a modern economy. "Let's hope the wind does what we hope it will....."
Why would Belgium and Holland wish to import abundant North Sea wind power? Are they not, as countries on the North Sea, generating their own?