Ep. 191 Andrew Lindsey, Chief Executive Officer at Flexnode | Data Center Go-to-Market Podcast

I, I love this because I actually just went through this yesterday, and it was a clear reminder that this is still in the space.

And, and I'm not saying that this is an impossible go-to-market strategy, but I'd, I wouldn't take it on anytime soon.

Um, it's the retrofit and the adaptive reuse go-to-market strategy.

Um, so you know, a, a big focus right now is, hey, are there buildings out there that have stranded power capacity, or better yet, even legacy data centers that have stranded power capacity that can be adapted f- for, upgraded for AI infrastructure? Um, and again, it's not impossible, um, but the economics can oftentimes be horrid.

Um, and I'll give three specific reasons why, uh, it can go sideways.

The first is understanding the power and the feasibility of delivering that power to that building, right? So just because a building has a, a, a substation access and a line or multiple lines to, uh, power doesn't necessarily mean that that building has the ability to take advantage of that power appropriately or put that power where it needs to be to effectively deliver the in, you know, to e- effectively activate the, and operate the infrastructure.

The, the, the sec- the second thing is, uh, related to then what happens if you've confirmed at a high level that you have power, but you don't necessarily know how you're gonna go about delivering that process of, you know, adapting the building.

The first thing that you'll find is that in the discovery process for many buildings that were delivered pre-five years ago, like no joke, five, 10 years ago, uh, you'll find that there's, uh, very either siloed information or isolated information about the facility, the operations and maintenance of the facility, all of the services, repairs, replacements that have been done to that facility.

Um, and as a result, there may be either one set of documents that has never been touched or there are several different set of documents that have different iterations.

So there's, you know, 10 different sets with five different versions, and they're hard to keep track of.

What that means though is that that discovery process takes months.

Months.

And unless you want to start peeling back walls in a building that you might not own yet, uh, to verify that a pipe might be there or, uh, that that structural load has actually been displaced as a result of the upgrade that happened two years ago or five years ago or 30 years ago, right? And then going from there, though, it's about then taking that information to a design build team and saying, "Okay- We have an understanding of what's in the facility.

We've verified that everything's in the facility this wa- wa- this way, that way, the other.

Here's how we wanna go about doing the proje- project.

Oftentimes those teams will be like, "Great.

Love that you did your homework.

We're gonna do the exact same thing because we don't want anybody to die on this job." So now you're starting again [laughs] in something that you've, you know, already gone about as a result of buying the, the building.

Now you're gonna do it for the project, right? So that's the second major issue of that adaptive reuse or upgrade.

The third is, who owns those upgrades at the end of the day? Especially if you've upgraded the facility for liquid cooling, where the customer requirements, if you don't own those GPUs, may have a unique set of technical requirements, uh, to effectively operate as an interface between the building and the IT.

What that realistically means, though, is that there's a pretty significant tenant improvement budget that needs to be accounted for, where if you're the tenant and you're asking the owner, right, they're basically saying, "That's entirely yours, and when you leave, it's entirely mine." If you're the building owner, you're thinking the exact same way from a flip perspective.

So w- major lesson learned, go to market, retrofit, adaptive reuse, huge headache.

Um, I wouldn't do it again in the near term.

What's interesting about that old habit die hard kind of thing is I personally experienced that with a huge amount of the colo operators that I was working with in the 2010s, where there was this hand-me-down [laughs] kind of idea-

Yeah.