Good morning and welcome to another episode of True Talks. Very excited today to have Matt
Caldwell with me. Matt, welcome to the program.

Thanks for having me, Eddie. Appreciate your time, bud. Thanks as always.

Matt, we have a global audience. Why don’t you spend just a few minutes and give folks a bit of
your background and where this data center journey has taken you?

Sure. So I’ve been in the data center space for over 20 years, graduated from URI, found myself at APC, American Power
Conversion, which was acquired by Schneider Electric.

So that brought me out to the West Coast. I managed the government enterprise accounts for
Schneider while getting my MBA from UCI, took that and went over to Siemens to manage
Hewlett Packard globally, did a brief stint at Alberio Energy before going to train for four years.
And for the past year and a half, I’ve been with Hyper, which is a manufacturer of low voltage
electrical equipment.

What a journey, man. Some amazing companies along the way. I’ve always admired the work
that Schneider does around the world.

So Matt, thanks for that. That’s super helpful. Matt, I wouldn’t even know where to start with this
data center conversation, but it’s moving fast.

It’s touching everyone, no matter where you live on the planet. You have some information about
data centers, but maybe a great place to start is, what do you see right now? What is happening
today in this incredible explosive market? Well, I think people are starting to wake up to data
centers. We’ve been in this space long enough to remember when people didn’t know what data
centers were.

And going to a family gathering for Easter or Christmas or something and having to explain for
the 10th time what a data center is to my relatives. And now in the past couple of years, they’re
coming to me and being like, oh my God, you’re in data centers. That must be super exciting,
right? So we’re definitely seeing data centers go mainstream.

And in conjunction with that, I think we saw this gradual rise over time of enterprise data centers
getting bigger and bigger and bigger. And eventually Amazon figured out that, hey, we can lease
out some spare capacity here and do some cloud computing. And we see this massive wave of
enterprise workloads going to the cloud.

And that in and of itself would be an explosive growth opportunity. But a couple of years ago we
landed on AI and now all of a sudden we blew the top off the data center market in parallel to this
exponential growth of migrating workloads to the cloud. Now we have AI on top of that with all the
training that’s going on for large language models and the inferences that we’re going to see
going forward is just an explosive time for the data center industry.

I don’t see it slowing down anytime soon. And it’s a good spot to be with us riding this wave. Matt,
isn’t it exciting that now when you’re at those family gatherings, they actually understand what
you do.

I have a similar experience. I mean, when I worked for a big civil engineering company for my first
25 years, you could point at big infrastructure and say, yeah, we did that, right? But when you’re
in technology, sometimes it’s that black box. What is it you actually do? Uncle Andy, you get that
question all the time.

So that has to be exciting for you. Yeah, and I feel like people at the same time know too much
and know too little, right? They’re very aware that data centers are being built in their community.

They’re very aware that, oh, there’s power, water usage associated with this.

There’s going to be some construction disruption going on. But at the same time, they’re not
really aware of what’s happening within the four walls of the building. Few people have ever
stepped foot into a hot aisle in the white space or a mechanical yard or understand that what’s
going on there is powering their apps, their AI engines, whatever technology they may be using.

They’re just under the assumption like, hey, that’s taking place on my phone or my laptop. They’re
not aware that it’s some server in a warehouse in Northern Virginia that’s actually doing that work.

Yeah, that’s fantastic.

Matt, if we think about this massive build, right? I mean, it’s in every headline. It’s all over local
community news. In your years of experience, what is the biggest change you’re seeing in that
build cycle?

Yeah, so I think early on, a data center construction project was a once in a fiveyear, a 10-year opportunity for a Fortune 100 company.

And you needed specialized resources to design a one-off box that you’re going to be using for
the next 30 years. What we’re seeing today is just this massive explosion of builds. And those
who are really building at scale are creating a repeatable model that they can do over and over
and over again.

And I think if you look back at it, what we’re going through with data centers right now could be
analogous to what the automobiles were doing back in the 1900s, where, yeah, there were
automobiles, but they were super expensive. They were kind of one-off. You knew who built that
engine because they signed off on it.

And it was a little bit of a, everyone had a little unique characteristic to it. And then Henry Ford
came along and said, no, we’re going to do it exactly the same way. It’s going to be an assembly
line.

We’re going to have a repeatable model and more guarantee on the quality and the speed and
the scale that you could build them out. I think we’re seeing a lot of that similarities today, where
when you talk to a data center developer, they’re not building one data center. They’re building a
massive gigawatt campus, and there’s going to be buildings as part of that.

And then they’re building another one that’s a mirror image. So they can use the same designs.
They can use the same parts for one data center building as part of that campus.

And then that campus is part of a broader strategy for data center builds. Yeah, that’s amazing. I
mean, when I hear you say that, some words that pop out for me is, industrialization of data
centers and then becoming more systems versus just one-off projects.

Matt, today, if you think about that delivery method, where does that break down today? I mean,
every time an industry goes through a massive rework, a revolution, what are some of the
breakdowns in that model today?

I think we’re just moving so fast right now that there’s not
necessarily the time to put the systems and the modules in place. And people are still using a lot
of the build strategies that they were using a decade ago today. And I think the companies that
are getting out ahead of this wave are adopting systems and modular approaches, especially
when you talk about the AI builds that are going on in West Texas or North Dakota, out in the
middle of nowhere, because latency is not as important.

And they can get cheaper power off the grid somewhere out there. I think what we’re seeing there
is that they’re adopting a little bit faster because there’s more pressure to do it in those types of
environments because labor is scarce. There’s not a lot of patience for rework in those sorts of
applications.

So they’re adopting that modular approach of, we’ll take a two or three megawatt block of power
and we’ll just buy X number of them and rinse and repeat over and over again. I think that
approach is allowing the industry to build at this gigawatt scale that we’re experiencing today.

That’s fantastic.

Speed is everything, right? You’ve got to have the facility up and running. There’s contracts,
there’s commitments. But how much does that cost control really matter at the program level? To
me, that’s something I always think through.

If you think about schedule scope and cost, what’s the impact? Is the focus more on speed,
getting it up, getting it running and cost is secondary or is that becoming important part of this
program development?

Yeah, you got to look at it very holistically, right? When you talk aboutcost, are you talking about component cost, the program cost or the overall total cost of ownership?

And I think when you look at the aggressive timelines that these data center
developers are committing to for their tenants, there’s a cost associated with not being able to hit
those delivery dates.

So when you look at the component cost that goes into that program cost
that goes into like, okay, what’s the total cost of ownership for this project?

It’s, you know, yes, price is a consideration, but the ability to deliver with speed and scale is really what’s driving the
market right now. So you need to be in the ballpark from a component level on price, but really
the speed and quality is going to be more important because if you’re the lowest cost, but you
miss that delivery date or the piece of equipment arrives on site and there’s something that needs
to be done that’s going to delay the project on site, there’s a higher cost to that than just the initial
purchase price.

And I think that as an industry, we are moving away from that design bid type of model where,
hey, we’re just going to blast out to the industry or specs and then, you know, choose on price or,
you know, relationships or whatever else. We’re moving into a world where those partnerships
need to be in place before the project is taking place. And I think at least the more savvy
procurement departments that I’ve seen know who their partners are before they start putting a
shovel in the ground.

And I’m not saying they’re sole sourcing everything, but they’ve got three different generator
manufacturers that they know they can rely on. They’ve got, you know, three suppliers for higher
medium voltage. They’ve got partners in place for low voltage and they have a pretty good handle
on the quality and the speed and the scale that those partners are able to deliver at.

And they’re just going through and being like, okay, you know, for this project, can you deliver,
you know, the initial building? But also if we’re going to standardize on you for the initial building,
can you supply us for this portion of the project and the overall campus? And those partnership
discussions are becoming more and more a part of this ecosystem we call, you know, the data
center industry.

Yeah, Matt, really helpful. And I think that, you know, if I think back, you know,
the developer has, you know, obviously, you know, speed, scale and, you know, just impact on
supply chain.

But if I’m an owner, you know, if I’m one of the big four from an owner and I’m building these
facilities myself, you know, you gotta pay attention to cost, right? Because you’ve got
commitments, you know, you’ve got lines of credit, you’ve got debt that you’re, you know,
accumulating. So I think there’s really that, it’s really a great bifurcation. Owners from a program
level have a different view than a developer that has to have the facility up and running and
carrying, you know, carrying data.

That’s a really good distinction. I appreciate that. Yeah, and when you look at the bigger guys,
you know, not the NeoClouds or the Wholesale Colos that are building for those, you know,
hyperscalers, but the hyperscalers themselves, the big logos that everybody would recognize,
they’re in a unique position because they’re looking, the size of the capacity that they’re buying
for, they’re entering into capacity agreements to the largest OEMs in the industry.

And they’re at a level where they’re looking to actually make investments in their suppliers. So
they’ll enter into a three or five year agreement with an OEM for billions of dollars of equipment to
guarantee that capacity going forward. And that’s a little bit different than a mid-tier Colo.

Some of the larger Colos are doing the same thing and basically just warehousing equipment for
their future builds. But if you’re a mid-tier Colo and you don’t have that level of buying power,
that’s where those more tactical relationships are going to be more of a priority for you than those
multi-billion dollar capacity agreements with the big OEM suppliers on electrical and mechanical
equipment. Yeah, that’s really helpful.

But I think what gets us excited at Contruent is that, you know, when we look at, you know,
chasing a program, a data center program, where people are starting to want to industrialize that,
the building’s important, right? But it’s supporting infrastructure. Today, we can help companies
deliver nuclear power plants. We can help companies build roads, bridges, highways, and rail.

So when we look at a data center project, we think about the building and the capacity that it
needs, but more importantly, that surrounding ecosystem from water to power. And that’s really
exciting. I mean, if you think about the size and scale of the total infrastructure needed to support
that facility, what’s your feedback on that? What’s some of the insights that you have?

Yeah, that’s a good question because I think five, seven years ago, we took a lot of that for granted,
right? There was a time where if you needed power from the utility, it wasn’t a question of if you
could get it, it’s just how quickly can they build the substation and how fast can we get that
transmission line.And it was a given that it was going to happen. It was just a question of what utility you were
working with and what your relationship was there. But it was a given that you could get power if
you needed it.

The scale at which we’re building data centers today, we’ve pretty much outstripped the supply of
the utility grid in North America. We’ve gotten to a point where a lot of data center developers are
looking at behind the meter onsite generation. Nuclear is definitely the next step in the
progression of data center development.

And the reason is, I mean, like I said, the utility grid’s pretty much tapped out. We’re tapping into
natural gas and onsite turbines to generate power, but we got to move beyond that pretty quickly.

And the idea of SMRs to power data centers, that’s a given. It’s just a matter of how quickly we can pivot to that. And I think there’s some regulatory hurdles that building on federal land is going to potentially bridge a little bit of that gap to nuclear power
for data centers. But yeah, it’s a different world that we’re living in now with the constraints on
power generation, water usage, land and fiber, just having that intersection of having the land and
the fiber connection somewhere where we can get power and use the water or be cognizant of
the water usage. So all those things create an environment where not every site is a viable data center site.
There’s a lot of times I enter into a conversation with somebody, they’re picking my brain. My
uncle has 100 acres in Mississippi. I think it’s a good data center site. And it’s like, I would question that. I’d say for every 10 that I
see, maybe one ends up being a viable site, but it doesn’t stop people from thinking they got one
there.

Yeah, for sure. Just a couple of points there. In the little community that I live in, there was a coal
fire power plant, one of early, early developments. And it’s now being converted to gas. And that’s creating jobs. That’s creating power for a huge data center that happens to be going right along the Ohio River. The other side of that, which is interesting for me, is there’s a wonderful road racetrack about 15 minutes from my house. The place is just amazing. It brings in various car clubs and you can go
out there and race four or five days a week. But the owner sold it to a major data center development. And it was one of those impacts that it
was like, wow, I get it. Guy has a lot of land, but obviously it was at the right location. It had access to power. It had access to good connectivity. But those are two that literally, both of those sites are maybe 10, 15 minutes from my home, which back to the human side of this. What do people get wrong about this explosive market of data centers, Matt?

Yeah, so to kind of tie into what you were just saying, I think adaptive reuse using existing infrastructure to build data
centers is good. I don’t necessarily condone converting a racetrack into one because then it
upsets the community because they’re like, hey, we used to have a racetrack. Now we’ve got a data center.
But a coal power plant or an industrial facility that you can convert into a data center, now you’re
using stranded capacity, stranded asset to build a data center. There’s not only some cost
benefits to doing that, but from a sustainability and a carbon footprint, you’re not building a new
facility. You’re reusing it.

And we should try to do that as much as possible. But the broader question, what are people
getting wrong about data centers? It goes back to, I think, the lack of knowledge and education
about what data centers actually do. I think that they at face value just see it as, hey, there’s this
big gray box in my neighborhood that replaced a racetrack or whatever.

And they’re not really understanding what is happening within there and the advancements that
are going on. I think there’s a knee-jerk reaction both to the building of data centers and AI in
general as a threat to the community, a threat to my job, a threat to really our society as a whole.

And I think that there’s a good amount of pushback to any sort of new technology.
But over the history of humanity, technology has taken time to adapt, whereas now we’re doing
this AI revolution in a matter of years that maybe would have taken decades or a century
previously. So I think that that pushback is understandable. But I think as an industry, we need to
go in and educate on the benefits of data centers, what’s going on within the four walls, what
benefits you’re going to see in your daily job or your daily life that is going to come from these AI
agents that are going to help you do daily tasks.

And yes, there will be some jobs that are displaced, but there’s also going to be jobs that are
going to be created. And there’s going to be whole industries that didn’t exist before and services
that wouldn’t exist without AI. And I’ve got the best example I have of that is listening to an
interview with Mark Zuckerberg, and he used a call center for Facebook as an example, right? So
there are five, seven billion Facebook users, and it doesn’t make sense to have a call center
because you would need millions of people to man a call center for Facebook.
But if you use AI, AI could answer 99.8% of those questions. The 0.02% that AI could not answer,
you could staff a small call center to handle those calls. And then as those questions get
answered, it just feeds back into the algorithm and those questions will be again answered
automatically in the future.

So a call center for Facebook could not exist without AI. And I think we’re going to see more and
more applications for new industries, new services that just didn’t exist before we had this tool at
our disposal. That’s fantastic, Matt.

Let’s double click on that just a second because Matt, as I kind of at the end of my career here,
well, the last 25%, I do spend a lot of time with young individuals and always the question that
they ask is, where is the next big job market? And for me, if we can think about, in two pieces, if
we can think about what type of technical jobs? I spend a lot of time making sure that people
understand the benefits of being an electrician, being a plumber, being a pipe fitter. I mean, that’s
so important to this. Matt, from the labor community, right, that next generation, where are the
biggest gaps today in that space?

Yeah, so someone with a, you know, 18-year-old freshman in college right now having this conversation and this thought experiment in real time, right? So back to your point about, you know, electricians and plumbers and the people turning wrenches out there.
If you are in high school right now and you have any sort of technical aptitude for that, 100% go
into it. Those jobs are going to be in such high demand and they are in such high demand. I’ll
give you, it’s been a couple of years, but I’ll quote Ben Stewart at NTT.
He got up on a stage at a DICE event and he said that they needed to hire, just based on their
current pipeline, they would have to hire one person per day for the next three years just to keep
up with what they had in their pipeline. And he said they were currently on track for one person
per week. So they were way behind schedule.

And he gave three ways at which we can close that gap, right? So the first idea that he had was,
give me your high school graduate with some sort of technical aptitude. And within four years,
they’ll be making six figures with zero debt as opposed to going to college. The other was
opening up the talent pool beyond just our traditional myopic view of like, here’s people who are
in the data center space, pull people in from other industries, advocate for women to come into
the data center industry, look for military veterans, those sorts of talent pools that aren’t
traditionally tapped into.And then the third was technology, right? If we’re short on labor, how can we do more with less?

How can we automate things? How can we use AI to augment the capabilities of our existing
workers and new workers coming into the workforce? Move back to my previous point of like how
I’m trying to sell my daughter on how to get into the data center industry. I’m like, kid, like this is
the hottest industry in our lifetime. You have connection through me and to my network.
I can give you an introduction to anybody that you need to. And her biggest objection is, just the
stereotypes that we’ve already touched on, right? Data centers are bad. Why are data centers
bad? Well, they use electricity, they use water, they’re killing the polar bears, all that sort of stuff.
So my advice to her is, be that advocate for sustainability in the data center space. It’s a home for
people like you in the data center world, which is be that resource within a data center company
developer that is saying, hey, is there a better way of doing this? How can we do it with the least
amount of environmental footprint, right? And I’ve introduced her to Connie over at Crane Data
Centers and others in the industry that are champions for sustainability and for the younger
workforce. So I’m doing my part as a parent to try to steer her in the right direction.
So we’ll see how far she gets. I love that story. And I really appreciate you sharing that.
And it’s just, it’s key. The trades are back. The trades are back bigger and stronger.
If you think about building the facility and running and maintaining the facility, I mean, that to me,
that’s the one truth that I want to get out in this is, it has amazing impacts to everyone, like you
said, that has an aptitude for a technical trade or even if you’re in computer science and
development and software development, it’s just the sky is absolutely the limit.

In the 42 years I’ve been in business, Matt, I can’t think of a time that I’ve been this excited about a technology
that’s going to just change the way that we all live and work. So Matt, coming down to the end
here, I got a couple more I really do want to get from you. What will separate the winners over the next few years in data center development? What are the things that you see that people are doing right? They’re just going to set them apart.

Yeah, and I think that goes back to the speed, scale and quality, right? And I think in order to do that, I think
you had said industrializing the process, right? Just go from, hey, you know, every data center
we’re going to design and build differently to just a very standardized approach. And I think that’s
going to be a partnership between the developers and the manufacturers and the design
influencers in this space.

It’s a collective effort here to move from that one off data center every five years to just a very
standardized approach. And I think what we’re going to see is, you know, a movement from, hey,
I can supply one piece of electrical equipment, another piece of electrical equipment, and then
you can assemble it on site or, you know, install it on site. Same thing on the mechanical side.
I think what we’re going to move to is, OK, I need, you know, 10 two megawatt blocks of power. I
need this mechanical configuration. OK, so I need five of those and, you know, three of these.
We’re going to just ship them on site for this phase of the project. And then when they arrive on
site, it’s just basically plug and play. You know, that’s going to that modular approach to several
things.

One, from a speed standpoint, if you just have a standardized block and you’re just grabbing a
few of them off the shelf, there’s not the same lead time expectation, right? We don’t have to be
like, all right, well, what’s the transformer that goes into this? What’s the, you know, breaker
configuration that goes into this? Whatever. It’s just, hey, we’re going to design around a
standardized block. And in order to design around a standardized block, you need your design
influencer, your engineer to build the design around that.

And you need the developer to adopt that modular mindset. But going into that, you’re going tohave faster speed because you’re just grabbing those blocks off a shelf. You’re not configuring them ahead of time.

You’re going to have the ability to scale because we’re just building these things in a repeatable
process. So you can just grab as many as you need. And then from a quality perspective, if
you’re doing the same thing over and over back to the assembly line, Henry Ford analogy, you’re
doing the same thing over and over again in a controlled environment.
There’s the likelihood that you’re going to have higher quality than a one-off system where
everything is just configured one time and put together and then, you know, shipped out to the
field and they’re, you know, doing some of the install work in the field. I’d rather have somebody
in a factory that’s doing it over and over again as opposed to the guy that’s, you know, just
learning it that first time following a manual being like, OK, well, this goes here. OK, well, I want
the guy that’s done this 12 times that day and knows exactly where it’s going, not the guy that’s
just trying to figure out how the pieces fit together.

That’s fantastic. Matt, kind of where that lands with me. I mean, first of all, it’s been a great
discussion and really appreciate your time.
But data centers are becoming more systems at scale. I mean, that that should get people super
excited. I know that’s why we have a complete go to market motion around that.
The real challenge, it sounds like, is coordination and then executing on that coordination is so
important. And then the next phase, really bringing more control and visibility into delivering, you
know, these these massive facilities all over the world. I mean, we’re super focused right now in
the US and I can’t imagine, you know, in other parts of the world how, you know, how big these
this market is.
So super great. Matt, I always like to have my guests kind of leave with the last word. Anything
you want to end with? Anything you’re passionate about? Anything you want to kind of deliver in
this in this last few seconds we have? No, I think that was great.
I think you hit the nail on the head as far as, you know, the systems approach to data center
development. And I