Infrastructure – surfing planks in puddles

Do you ever worry that you’re perched on a surfboard half a mile out to sea waiting to catch a tsunami… but in reality, you’re teetering on a plank in a puddle?

If that feels familiar, you’re likely working in the infrastructure sector.

All summer we’ve been hearing that it’s kicking off come September. Actually, we hear that every summer and – for blindingly obvious reasons – those who share these nuggets of wisdom… tend to be right.

There are stacks of M&A deals in train and as one old mate (joyfully mixing the above metaphor) says: “It’s like one of those cowboy movies. Last night, we could hear the drums from the other side of the mountain and we’re waiting to be slammed this morning. But right now, it’s quiet… too quiet.”

Well, that’s all fair and good. There’s a lot of M&A on the radar. As to whether the assets fetch the numbers vendors are seeking to achieve, that’s a very different matter.

It looks a shade gloomy when you take on board one investment banker’s doleful rumination from earlier this year that there was a 20% success rate on processes… which is leading a rise in continuation funds… and continuations of continuation funds.

So, we cower in circled wagons, pondering the wisdom of dynamite crate barricades, and look to the unbridled enthusiasm of the greenfield sector.

Blimey, that’s not a whole bunch better.

Offshore wind has created its very own dunkelflaute that’s sucking the wind out of its sails, PPP’s on life support, ESG’s being largely sidelined, while the infra community’s throwing its (not inconsiderable) weight behind data centres and the energy required to power AI.

Now, I appreciate this is kinda bucking the trend, but is AI all it’s cracked up to be?

Technology risk has never been a comfortable bed fellow for infrastructure. It just moves too fast for a sector that pitches its stall at long-term investments (even though assets sit in 10-year maturity funds) and long-tenor debt.

This is why EV charging as an investment is such a concern. Last year (2024) the European Environment Agency published statistics that the average passenger car across the EU was 12.3 years old. In the UK, those stats are almost 10 years while Spain weighs in at 13-and-a-half and Greece is a whopping 17.

Now there are any number of reasons for those stats – ranging from economic to cultural, and everything in between – but I’m holding off upgrading until technology settles down.

And, frankly, a 15-year-old Merc with fewer than 50,000 miles on the clock will do us nicely for another 5 years, thank you very much… even if the dog’s doing her best to destroy the interior.

Genuinely artificial “intelligence”

Talking to folk around the industry this week, there are 3 schools of thought on AI in infra, and its future – there are true believers, Doubting Thomases and bubble burst bandits.

True believers are of the view that AI will take over (some muttering that journalists are the first against the wall) as they crack out riding crops to horse whip juniors out of the building… thereby creating succession issues… but who cares as it’ll see them to retirement.

The Doubting Thomas school reckons it will impact infrastructure, but not to the extent some people believe. In the same breath, they say it will impact infra more than some folk think. So, that’s the middle ground well-and-truly occupied.

Fans of the bubble burst scenario stare with glee in anticipation of the whole darned stack of cards hitting the table, with a lot of renewable energy going begging. As one source quips: “If data centres have been overhyped, we may reach Net Zero sooner than expected!”

But – given the amount of airspace being given to AI, data centres and the electricity required to power and cool these monsters – it’s no joking matter.

As one infra fund veteran of the debt persuasion says: “For infrastructure, everyone’s a lot more cautious on future AI usage in data centres as – frankly – no one has a clue.”

And that’s the truth of it. Nobody does have a clue… which is worrying in a financing community where the default is to stick with known knowns, shunning the funky.

Another fund source, but this one with a leaning towards equity, says: “AI has massive applications, but we have yet to figure out where it will be really applicable… whereas the current assumptions sort of imply 100% application across the board.

“It has wider implications too across infrastructure assets classes. Will we need the Bakerloo extension and more train networks if white collar jobs get disintermediated.

“There are a whole host of second order implications – mostly negative – that I don’t think are being ‘valued’.”

And then there’s the banker who takes discussion down the crypto line which, as the lender says “was supposed to hoover up gargantuan energy but ended up being – in relative terms – a damp squib”.

The source adds: “I'm in the contrarian camp – that AI will have its uses, but will not be the ‘everything, everywhere, all at once’ prophesy being pushed out by big tech. It's in their interests to do so – they need the next productivity leap that spreadsheets gave us in the 1980s and 1990s.

“I reckon it does become additive to energy demand, but probably in the order of another industry like aviation in its consumption. Then we also contend with general energy demand growth over time. More energy is needed, but AI is unlikely to dwarf everything else – additive yes, but not all encompassing.”

And then you have the infra deep thinker: “In my opinion the genie is out of the bottle. Social capital was initially more committed here than actual capital. But the hype was necessary to get all wealth funds and asset classes upended and panicked into throwing their cash at this brave new world.

“Classic FOMO behaviour by the dumb realities of money. The social capital just needed to be caught up in the hype and the then cash flows to the promise – not the business case.”

The source continues: “These projects need so much resource – water, energy, cash, intellectual focus – and right now real stupidity sits ahead of artificial intelligence. It is what it is, at least for now.

“The hopeful note is to return to what humanity is best at. History is full of us eventually finding something worthwhile out of a panacea of misery. Something good will come of this eventually.

“But profit and prophecy are currently in make-believe mode, and hubris will be our painful reality first. Then we can start thinking again… because right now that seems to be what most have forgotten!”

And then there’s the banker whose search history would benefit from closer inspection: “Massive data centres close to renewable energy projects, and close to cities with available water to cool things down, with top notch security and so on… all to run porn sites and ChatGPT?”

The companies driving data centres have deep pockets and so are in the best position to take this risk, but the hype that is driving unthinking commitment to a – relatively – unproven thesis.

The upside should all go to hell in a handcart, nations take a step closer to Net Zero with all these lovely wind farms and solar parks – maybe a few SMRs if Microsoft gets its way – and the world’s a better place.

Meanwhile, those at risk hold on to their pen-pushing / keyboard-prodding jobs.