Is the real estate market ignoring red flags when it comes to data centres?

Real estate investing tends to come with a familiar checklist: What's the yield? Where are interest rates headed? How does it stack up against the index?

It's a reasonable starting point, but it can mean investors end up chasing whatever's hot rather than where the value is.

Global real estate is trading at some of the widest discounts to replacement cost in years, even as the AI capex boom catapults data centre values through the roof.

Chris Bedingfield of Quay Global Investors runs a different playbook, built on two ideas. The first is replacement cost - what would it cost to build this asset today? The second is needs based - whether people and businesses can really do without the property in question.

In this interview, he talks about how these two filters shape nearly every call his fund makes - from why property developers are excluded from the portfolio altogether, to why his fund holds a smaller data centre position than the benchmark, to where he thinks the next few years of returns are going to come from. 

Commodity vs franchise

Bedingfield splits real estate into two camps. "Commodity" property (office towers, industrial sheds, self-storage) is relatively simpler and cheaper to replicate, which means it has less pricing power. 

"Franchise" property is the opposite. These are assets that can't easily be reproduced because of scale, planning restrictions, or how dependent tenants are on them. Chadstone in Melbourne is an example. "You couldn't build another Chadstone across the road," he says. "The planning, the tenants, the scale, you just can't do it."

This difference shapes how Bedingfield thinks about risk. For commodity assets, he wants to buy below replacement cost, because that's the level at which developers have no reason to add new supply, meaning less competition for existing owners.

Right now, he says that opportunity is everywhere. "I can't get out of bed in the morning without tripping over a REIT that's not trading at a 30% discount to replacement cost," he says. "There's value everywhere."

A supply crunch a few years out

Bedingfield says this mispricing is setting up the next cycle.

Construction costs have jumped since the pandemic while pricing across most of real estate has stayed soft. Very little is getting built right now outside data centres and industrial.

"Looking one, two, three years down the track, you are going to have a real crunch of lack of supply with ongoing demand," he says. "That means accelerating rents, that means accelerating capital values."

He points to Australian residential, where rents are being squeezed by exactly this dynamic. 

“That theme is happening in commercial real estate right around the world in a major way. And that'll continue to happen for some time, we think.”

The AI capex math doesn't quite add up yet

Given how much capital is being funnelled into AI infrastructure, the big question is when do the economics of this story stop making sense?

Estimates put global AI capital expenditure at around US$5.5 trillion by 2030. Bedingfield says that at a 10% return hurdle, that means the industry needs to generate roughly US$550 billion in total profit against total US corporate profit, both listed and unlisted, of about US$3.5 trillion. On its own, he says, that "doesn't feel like a stretch".

"The chips that they're using have a reasonably short depreciation life - Meta thinks the depreciation life of these chips is about five years, others think it might be one to two years," he says. “That depreciation charge could be in the trillions.” 

Considering the capital intensity of this tech cycle and subsequent depreciation charge, the “profit need is enormous”.

Then there's competition. Unlike Apple, Google or Meta, Bedingfield argues the AI models don't have an economic moat yet. "The switching costs between using AI models at the moment is zero," he says, pointing to reports that OpenAI has been cutting token prices to get their user numbers up.

“For us, there are definitely red flags in terms of the sustainability of this [AI] industry… and I think the jury is out whether the economic returns are there."

No developers

A strict exclusion from the Quay funds are property developers. Bedingfield explains:

“Real estate cycles are very long, but they're very predictable. Nothing trades above replacement costs forever and eventually prices reset and come back down again."

"I can't overstate this enough, the risk level in the development business is extremely high. As soon as you put a shovel in the ground as a developer, you're committed," he says. 

"You're spending money for the next two, three, four years... demand could change on a whim, but you're committed with that capital."

Data centre exposure

Quay's data centre exposure sits at around 3% of the portfolio, and has been trimmed through the year. While Bedingfield doesn’t doubt the demand story, he thinks the market has stopped questioning the pricing.

He describes touring a two-year-old data centre that was still only 70% leased, with a second one being built right next door with no tenant secured. 

 "If this were office sector and a building was open for two years and 70% leased, and a vacant one being built right next door, most real estate people would be running around with their hair on fire," he says. 

"But because it's data centres, we all seem to have lost our collective ability to think."

Where the value lies instead

Quay's highest-conviction holdings sit in retail, senior housing and self-storage - property sectors trading below replacement cost, with demand that holds up regardless of what the economy is doing.

Senior housing in the US is a standout for Bedingfield with a strong demographic tailwind running into housing starts sitting at 15-year lows.

He draws a comparison to the early 2000s, when real estate quietly compounded while the market chased the dot-com boom, then delivered close to 100% total returns in the years after that boom corrected.

His base case for the next few years is similar - double-digit total returns from real estate, coming not from the sectors dominating the headlines today, but from the ones most investors have stopped watching.