In recent months, the deployment of Artificial Intelligence (AI) technology has been pushed as a key way forward for economic growth, for example by Kier Starmer in the UK and Trump in the USA. See my earlier post. It’s certainly been seen as a big deal, by the USA especially, not least in terms of its likely impact on power demand. Globally, by 2035, IDTechEx forecasts that the continued growth of artificial intelligence will result in over 2000 TWh of energy being consumed by data centres. So with a vast and increasing amount of energy being used to churn through ever larger data bases, emission targets may be hard to achieve. Some US big data companies seemed to think that green sources would not be enough to deal with this and that a big expansion of nuclear would be necessary- either new Small Modular Reactors or revamped large plants.
In one of its scenarios, Goodman Sachs Research claimed that ‘85-90 GW of new nuclear capacity would be needed to meet all of the data centre power demand growth expected by 2030 (relative to 2023),’ but it noted that, actually, ‘well less than 10% will be available globally by 2030’. So, it concluded that ‘nuclear can’t meet all of the increased data-centre power needs’, and that ‘natural gas, renewables & battery technology will also have a role to play.’ Even so, nuclear would still be the main option.
Not everyone agreed with that conclusion. For example in his new comprehensive overview of the state if play in AI, BNEF founder Michael Liebreich said it was ‘unlikely to be powered by SMRs, fusion or space solar for the next few decades’. Instead it will be renewables. Trump would be unlikely to concur, although he says he would not apply any constraints on new energy technology for helping to push AI fast, but that includes coal plants!
However, all this policy thinking, speculation and debate now looks rather dated- following China’s release of a new high performance Deep Seek AI technology, which cost much less than its mainly US rivals and may use less energy per search. It’s early days yet, but it feels like a ‘sputnik’ moment, with the USA being outflanked technologically, this time not by a Russian satellite as in 1957, but by a Chinese super-AI system!
So, for now, all bets are off, and US technology markets have been in turmoil, with a new tech race off and running- and some bans on using Deep Seek in the USA! Clearly, assumption about growth in the US (and UK) being aided by AI dominance will have to be revisited. So too may assumptions about energy needs. For example, share values in Constellation Energy, the company behind the planned revival of the Three Mile Island nuclear plant to power AI, have already fallen 21%. So now we may not hear so much PR noise about the need for nuclear - renewables should suffice.
None of this means that AI will not happen- it may well become dominant. But it may be cheaper, and less energy demanding- and more Chinese! Much like renewables. In terms of what happens in the UK, there will no doubt be renewed efforts to stay in the AI race and to link this (and much else) to renewable expansion. So, there will still be some big issues to face, for example, concerning how the transition is managed regionally.
A recent Prospect Magazine Supplement had an interesting article by Ross Mudie from Icon which noted that ‘the green transition and AI will be transformative - but the question is whether their benefits will be distributed equitably’ . It warned that ‘both transitions could exacerbate existing inequalities rather than reduce them. Local UK economies that are already strong will be able to capitalise on these changes, but poorer parts of the country could lose out further. Without intentional, proactive policymaking, we risk repeating the mistakes of deindustrialisation and leaving more communities behind’.
Well that certainly maps on to current planning and infrastructure debates e.g. over the potential for growth that might follow Heathrow expansion or the development of the Oxford – Cambridge high tech enterprise corridor as proposed (again!) by Labour. The South East is rich and will probably do well from AI, but, unlike some of he relatively poorer and undeveloped areas in the North and West, maybe less so from renewables- given the environmental constraints. Although runways and motorways in the SE are eco-issues too! And local opposition to change is likely in most places, possibly made worse by Labours move to centralise some planning decisions and remove ‘infrastructure blockers’ via aggressive planning policies.
Mudie looks at complex regional issues like this and says that dealing with them ‘means place-by-place policy interventions, rather than acting as if one size fits all. It means targeting investment at places that need it. And it means ensuring workers displaced from struggling industries are reskilled’. As some of my earlier posts have indicated, progress is being made on this, for example in relation to green skills passports, and as some of the other articles in the new Prospect supplement show, other idea and plans are emerging, for example from the newly created Skills England initiative. However, there is a long way to go with green skills in all areas and sectors, and, as for AI, we are still mostly at the starting gate, with, in any case, not everyone on board. Some see it as dangerous for jobs – and even for humanity.
There are also big ecological and social uncertainties about AI-led growth, or indeed any sort of growth, if it just means more of the same sort of unhealthy consumerism. Some alternative economic frameworks have been proposed over the years, as I explored in an earlier post. While full-on de-growth may be too radical for some, a more sustainable approach to consumption and production, along with structural changes to the economic system, should be possible, and indeed is seriously needed. And it’s just possible that AI could actually help us move in this direction- or is that naive techie utopianism?
Comments
Post a Comment