Prologis Hopes One Day AI Will Help Source Its Acquisitions

At the start of the year, industrial giant Prologis was making some bold predictions.

“Our projections take the bull case on interest rate cuts,” they wrote. “While the latest Federal Open Market Committee member projections from December show a median federal funds rate in 2024 in the mid-4% range, lower than 2023’s mid-5% range, we expect inflation to slow more quickly than anticipated.” The second part of their argument is that the 10-year Treasury yield would “dip below 4% in 2024, ahead of consensus views for around 4%.”

Instead, inflation has moved back up a bit in the first two months of the year and the famous saying about the weather in March has largely become in like a bull, out like a bear as the 10-year Treasury yield is as of last Friday 4.22%.

As for Prologis, finding deals for them has become much tougher because of macroeconomic factors. And asked whether artificial intelligence could help them, the answer is maybe eventually, but the technology that has been doing the yeoman’s work has been human-led data analysis.

Nina Trentmann at Bloomberg recently spoke with Prologis Chief Financial Officer Tim Arndt and Managing Director Chris Caton.

“There’s indeed a dwindling amount of such portfolios out there,” Arndt said. “There are privately owned portfolios, not only in the US but elsewhere in the world that we would have keen interest in. But that’s where you need a willing seller. The balance sheet is capable and has a lot of capacity to do almost any deal. There’s no portfolio out there that we couldn’t take on quite easily. We’re always looking.”

Surprising/? No. CoStar noted earlier this month that industrial was on track to deliver the weakest first quarter performance in a decade. Among the factors slowing absorption are third-party logistics companies and retailers that beefed up demand early in pandemic to meet e-commerce and home delivery demands. But as things have normalized, they’ve needed less space and have closed excess distribution centers and put them up for sublease.

One example of how AI — whatever of the many types is in question, because they vary a lot — might fit in is provided by Blackstone, which has been using it to build in-house tools to improve their investment process.

However, Caton told Bloomberg that Prologis has been using good analytics in their work. “For AI to really thrive, you need tons of observations,” he said. “We own 5,000 buildings, there’s 75,000 buildings globally that are really interesting. And so, the range of decision making for site selection is rather discreet in the scope of what it takes to train an AI model.”

He adds: “We aspire to have it say, “Look, the property on the right hand side of the street is going to be better than the one on the left hand side of the street.” We aren’t there yet though.”