4th December 2025: The Overlapping Cycles of Hydrogen and AI
This article appeared in slightly abridged form on LinkedIn where the author had been previously unaware of the character limit for posting.
Like many others, I was not totally surprised by the announcement that BP is abandoning its CCS and hydrogen plans at the Teesworks site, and that a hyperscale AI datacentre will be developed in its place. My first reaction was that a serious proposal had been supplanted by one less serious – but such reductive thinking doesn’t help matters, so I unpacked it further.
Many in the network will know me for my work supporting the development of hydrogen projects, all green ones, to be fair, and not blue as was proposed on Teesside.
In the face of some serious geopolitical headwinds in recent years, the oil and gas industry (to my perception at least) have leaned in to the notion of a post-climate aware world a little too willingly, almost as if it was seen as a bothersome fig-leaf which had become an encumbrance now that false modesty was no longer an issue.
BP’s withdrawal from Teesworks could be seen as part of that trend, but what is undeniable is that hydrogen is expensive and complex to develop and we (the UK) seem to have stalled short of a critical mass of projects to truly create an industry. I am a huge advocate of hydrogen where actual supply meets actual demand, and there are still some serious developers working in this space. Five years ago there was perhaps some less serious thinking and a far more speculative landscape for hydrogen development, and what one of my recent colleagues refers to as the beginning of the “hype cycle”. As we’ve seen, what followed was a massive rationalisation as the bubble shrank back to the size of the serious projects.
Economists and environmentalists alike fancy themselves as experts in cycles, and it struck me as emblematic that a hydrogen project during the “bust” part of the hype cycle is being bodily replaced on the same land parcel by an AI datacentre – surely on the “boom” part of theirs.
I definitely know more about hydrogen than I do about AI but can see clear serious and unserious applications. The valuation of AI developers is so overcooked that the talk about a bubble about to burst has become almost casual. Every second post on LinkedIn seems to be accompanied by some unfunny AI slop cartoon, AI seems to be spuriously shoehorned into every app and operating system and don’t start me on the really grim stuff you can see – and make.
But I’m well aware there are serious uses and it can shoulder the burden for many previously onerous tasks. I have a client who develop medical AI solutions and have empirical data to prove that their AI reduces emergency hospital admissions among vulnerable patient groups. Perhaps these are the sorts of serious uses that the industry will contract to if and when the AI bubble bursts, though I’m in no doubt the residual AI industry will still be huge.
I wish the developers of the Teesworks data centre every success – in fact I think I might have written some of the EIA within a couple of wrappers. I just hope it is going ahead for the right reasons and they aren’t left with a slightly warm portal frame warehouse looking for a new tenant in a couple of years.
I’ll leave this post up to see just how badly it ages!