Land for food versus AI and Solar Power…

Love him or hate him, Jeremy Clarkson and the Amazon ‘Clarkson’s Farm’ series have highlighted the reality of British Farming more than any other programme. Farming is tough. The hours, bureaucracy, and regulations are horrendous, and making a meagre profit is an   unrelenting challenge. Fifty years ago, the UK imported ~35% of its food and today it’s around 40%. For an island with prime agricultural land, one might expect the country to want to maximise the use of every inch of it to achieve greater food self-sufficiency, especially given lessons from the COVID pandemic, the Ukraine and Middle East conflicts, and other geopolitical tensions. Sadly, that doesn’t seem to be the case. Prime agricultural land, as the article here illustrates, is under attack. Rather than being used for food, it’s being converted into huge solar farms because ‘the only way to have energy security is if we take a pro-growth approach to building more clean energy in Britain’. That’s ideological balderdash, of course, from a government minister who’s just overruled the recommendations of his own inspectors to approve the use of prime agricultural land for a massive solar farm.

Fifty years ago, ~95% of our electricity was generated using coal and oil, half of which was imported. Today the UK mix of electricity generation is very different, as can be seen on the live energy dashboard here. At the time of writing, solar is 27%, wind 25%, nuclear 8% and 23% is imported across interconnectors with Europe. Our electricity generation self-sufficiency has improved way more than our food self-sufficiency over the last 50 years, so why use prime agricultural land for even more capacity? Why not use brownfield sites or car parks? The answer, the Badger senses, may rest in something JD Vance, the US Vice President, said recently, namely ‘the UK has been failed by its leadership for a long time’. An uncomfortable truth, perhaps, particularly when it comes to the use of prime agricultural land.

The Badger’s not a farmer, so why this focus on the use of agricultural land? Because our countryside, and everything in it, is being eroded to develop huge solar farms and huge electricity and water hungry, AI data centres – see here, here, here, and here, for example. AI fever, anti-fossil fuel zealotry, and ideological dogma seem to trump the common sense of better food self-sufficiency and preservation of our ‘green and pleasant land’. In 1804, William Blake used this phrase in his poem, Jerusalem, to reflect a longing for a simpler, calmer country during the Industrial Revolution. This still resonates today with AI and electricity at the heart of a modern industrial revolution. Will the UK’s new Prime Minister prove JD Vance wrong, preserve the UK’s ‘green and pleasant land’, keep AI and ideological zealotry in check, and improve food self-sufficiency? Don’t hold your breath…

OpenAI pausing Stargate UK is hardly a surprise!

As widely reported (see here for example), OpenAI is pausing its multi-billion-dollar Stargate UK project. The project was first announced in September 2025 with the declared purpose of ensuring ‘OpenAI’s world leading AI models can run on local computing power in the UK, for UK – particularly for specialist use cases where jurisdiction matters. This will help power the UK’s future economy, boost its global competitiveness, and deliver on the countries national AI Opportunities Action Plan’. The UK government’s AI Opportunities Action Plan had been announced in January 2025 as a focus for ramping up AI adoption to boost economic growth, jobs, and improvements to people’s everyday lives. A year later, in January 2026, a seemingly positive  progress update was published. The government’s thus likely to be wringing its hands about OpenAI’s pause. Why? Because it puts a dent in the country’s desire to be an ‘AI superpower’, especially when the company asserts that regulation and high energy costs are obstacles. The Stargate UK pause, however, is hardly a surprise given that the holistic situation faced by OpenAI today is really no different to when the project was announced last September.

OpenAI announced the project on the date President Trump started his state visit to the UK. With tariffs as a backdrop, the pressure on the UK government to make the visit a success was huge, and a centrepiece during the visit was the signing of a technology partnership involving new investment and cooperation on AI. Domestically, the government needed this to promote its growth agenda, but a ‘technology partnership’ and tangible realities are different. Given the pressure for the visit to be a success, OpenAI’s Stargate UK announcement was part of an overall joint PR strategy – at least that’s what the Badger senses. At that time, the UK had some of the highest costs for electricity in the world, and that’s still the same today! If there’s one thing an aspirant AI superpower needs, it’s economically competitive electricity and so it can hardly be a surprise when a commercial company focused ‘on the business case and numbers’ decides to hold off further investment. Additionally, there’s uncertainty about changes to UK law to allow AI firms to train their systems using copyrighted works, ongoing investor anxiety about an AI bubble, the fact that OpenAI hasn’t delivered a profit yet and is forecast to make losses of ~$44 billion before becoming profitable in 2029, and that OpenAI is facing massive competition from Google (and others) which is raising significant questions about its future. All of these points were material when Stargate UK was announced 7 months ago, and they remain so today.

A sceptic could thus be excused for thinking that the project was driven by a geopolitical public relations necessity in the first place. For the Badger, with his instincts rattling from experience, it’s thus hardly a surprise that Stargate UK is paused…   

Nuclear Power for AI Data Centres…

According to the World Nuclear Industry Status Report, whose data can be explored visually here, there are 407 operational nuclear reactors currently generating electricity across the world. Of these, 94 are in the USA, 62 are in China, 57 are in France, and 34 are in Russia. The average age of the world’s operational reactors is 32.6 years, and they generate ~9% of global electricity. There are ~11,800 data centres worldwide with a rapidly growing proportion incorporating AI-specific infrastructure. Whereas traditional data centres require 10-15KW of electricity per rack, AI data centres need 40 – 250 KW per rack to support the heavy computational demand of AI models. So, where’s this extra electricity coming from? It’s a question brought into sharper focus by the conflict in the Middle East and its potential impact on the availability and price of gas which is used to generate ~20% of electricity globally.

All the major tech giants have been considering this question for some time. They want a reliable electricity supply and low emissions for their AI endeavours and are thus turning to nuclear power. For example, Microsoft wants to restart a Three Mile Island reactor mothballed in 2019, and Meta have signed a trio of nuclear deals  securing enough electricity to power ~5 million homes for its AI data centres. It takes some decades to build new, large-scale, nuclear reactors like those currently connected to electricity grids, and the surge in power demand for AI data centres is surpassing the planned new generation and transmission capacity. Amazon and most of the tech giants are thus keen to harness Small Modular Reactors (SMR) to sustain AI growth. SMRs are new with just two in the world currently operable. However, you’ll see from the World Nuclear Association’s SMR project tracker that we can expect many more to come on stream over the next decade.

Nuclear SMRs will thus be key providers of the power for the AI data centres needed to underpin this digital technology’s ever more rapid momentum. Is that a problem? No, provided there’s strict regulatory control before, during, and after SMRs are built and put into service, and that global institutions exist with real teeth to ensure that commercial organisations and nation states do not flout the necessary balance between AI self-interest, the greater good, and the proliferation of nuclear material. That may be a tall ask in a world which is full of conflict, extremism, and volatility, and is already embarked on a huge race for AI dominance. SMRs, however, are new and things may not go to plan. If SMR delays happen, then we may see AI momentum slow over the coming decade. Electricity, after all, is the blood of the digital world, and if there isn’t enough blood then things are bound to go awry…