EE#13 Britain's New Power Couple: AI & Energy
Jeremy Hunt could turn our gridlock to his advantage
When Westminster says grid it normally means Alastair Campbell’s No10 media grid to manage an increasingly complex media environment. But last week, Westminster turned to a more tangible but equally complex grid, one which Jeremy Hunt has at the heart of his new growth plan.
The electricity grid is typical of Britain’s stalled economy. Like our ageing railways, we built ours first, benefitted, but then failed to keep it up to date. How our energy system can handle changing supply from new variable, renewable, electricity, alongside a huge increase in demand is a major headache. National Grid says it will need to build more transmission than five times more in the next seven years than in the last thirty. It’s even harder being an island - we can’t rely on as much electricity imports as our continental neighbours.
The UK has the longest grid connection queue in Europe, with 200GW of power, totalling £200bn in investment, tapping their watch - four times current capacity. Some projects will wait until almost 2040 for a connection, five years later than when we should have decarbonised electricity. Meanwhile, we pay energy generators to switch off to prevent transmission lines overheating - so-called constraint costs. Companies are cancelling labs, data centres and much-needed family homes. Growth and net zero are in gridlock.
There is progress. The UK has the highest proportion globally of renewables already online. Ofgem, the regulator, is reducing the waiting time for a connection. Alongside decarbonisation, National Grid is also decentralising and digitalising (take note Westminster) all whilst making sure your oven still cooks dinner tonight. They are essentially changing an aeroplane’s engine mid-flight, whilst ground control is telling them to go full throttle.
Decarbonisation is worth the challenge. Greening power will improve our economy by virtue of doing the same things we do now, better. A heat pump produces the same heat as a gas boiler but for only a quarter of the energy. Less money spent on energy frees up investment for more productive uses.
Climate advocates point to industries created on route to net zero that would also boost UK GDP. That conversation often has a nostalgic tilt, hoping the UK builds stuff again, like in the good old days. It tends to miss what we are best at and sell most of abroad. Our greatest specialisation in clean tech is in services. UK comparative advantage is in energy management, not making cables.
Our ageing infrastructure has forced an upside. National Grid has gotten very good at using software to manage energy. This is also where some of the safest returns are emerging for green-minded investors, with companies like Blue Bear Capital stacking big numbers on clean-digital alone.
It’s not just energy where the UK uses new software to overcome ancient hardware - DeepMind’s AlphaGo, developed in the UK, was the first computer programme to beat a human at the several thousand years old game Go. Our AI advantage is real but is currently focused in science and research in similarly venerable universities.
Both AI and clean power are now hitting a classic UK economic problem, taking an early technological advantage and deploying it in the real economy, whether that’s ensuring everyone has a smart meter or some level of computational skills.
Thanks to Biden’s mega-spending or the EU’s regulatory superpowers, international competition is heating up. That makes maintaining a technological advantage and turning it into an economic advantage harder. Policymakers need to choose, but there are few better choices than where two of our current strengths overlap.
The UK should be using AI to accelerate and balance a zero-carbon grid. Decarbonisation means more components, more complex decisions, and more computational power required. But to AI that is just more data to feed on.
In practice this could mean more accurate predictions of energy supply and demand, bringing weather forecasts, consumer behaviour, and transmission closer together. It could also accelerate the development of Virtual Power Plants, allowing National Grid to combine a distributed set of energy resources - batteries, EVs, data centres - and draw energy from them to meet peak demand - as we do with gas now.
For consumers, AI could help you predict the most cost or energy-effective time to put a wash on, or even discharge your electric vehicle to sell energy back to suppliers. Ultimately, a better balance of demand and supply should lead to less energy required for the same outcomes - a cost-saving and productivity boost.
Given that our clean power targets are faster than other countries (2030 for Labour, 2035 for everyone else), this could be a big export opportunity - both the software and the engineers running it.
This is not a case of just doing some new tech and ta da sunlight uplands. There are challenges. The most immediate is technical. The grid is full of information asymmetries: producers, transmitters, and consumers all have their own data on consumption and behaviour, and struggle to speak to each other. Smart meter take-up is off track. Where different companies are involved there are collusion issues (though the Competition & Markets Authority is supporting businesses to work together on net zero).
Then there is the public. We know from the heat pump rollout that people like to feel in control of their energy. To give people that sense, some manufacturers now put buttons on that flash some lights but do little else. How would the public feel about AI controlling national infrastructure, or the energy in their home? Would they care if they didn’t know? Balancing is done by National Grid anyway, but I expect people would imagine an accountable person pressing buttons. Digitalisation could also increase cyber risk to critical infrastructure -a security concerns. This is not unique to energy, ethics and accountability are the same in other AI applications but are maybe more tangible than military drones.
Overcoming challenges means testing. Ofgem is good at regulatory sandboxes, a way for businesses to trial new ideas under the eye of a regulator, but these are usually product- not place-based. The grid is a physical thing. To trial different ways of running it we need to see real-world effects, whilst preventing the lights turning off across the whole country. We need to move quickly. Ofgem’s regulatory sandboxes last up to two years. If they started today that would only leave a decade to completely green the grid.
Ofgem and Whitehall should provide National Grid and businesses a safe space to prove the efficacy of AI in the grid. This needs to be somewhere close to renewable generation, with a university for research, and that uses lots of energy but not time-critically, like heavy industry. This is not dissimilar to government’s hydrogen village plan for Redcar.
Westminster has begun paying attention to how the grid has stalled the economy. Focusing on managing the problem will though only move us level with other countries. If we want to properly grow and compete, we need to go beyond. Using the UK’s existing expertise to solve and run a green grid, faster, means we can sell services to help other countries do the same. Gridlock, unlocked.
This piece as well as being something I’ve wanted to write for a while is also my entry TxP progress prize, a blogging contest on the question “Britain is stuck. How can we get it moving again?