AI is pushing data center growth, but labor is limited

As generative AI grows in popularity, the demand for data centers is skyrocketing. Clayco founder and executive chairman Bob Clark joins Market Domination Overtime to discuss this trend.

Clark describes the pace of data center construction as "insane" with "tremendous" demand behind it. He explains the two main drivers of this demand: inference, which he explains is "the digitization of everything," and AI, described as "the trainer model... the learning buildings."

Inference-driven data centers are often built in major cities to "stay close to the use," Clark notes. However, AI data centers can be constructed anywhere with a sufficient power supply.

A significant challenge in this boom is labor availability. Clark points out that with some projects requiring 800 to 1,000 workers, labor can become "a limiting factor" in data center construction.

For more expert insight and the latest market action, click here to watch this full episode of Market Domination Overtime.

This post was written by Angel Smith

Video Transcript

As the demand for a I grow, so does the demand for data centres that make it all possible.

Joining us Now.

Bob Clark, Clay co founder and executive chairman.

His company helps build those data centres among many, many other things.

Bob, it's good to see you.

Thanks for being here.

Thanks for having me.

So we have been talking a lot recently about a I, um As you might imagine, we've sort of been talking about how the story has been changing a little bit.

And there's still a lot of talk about data centres.

There's also, however, more and more questions about how a I is gonna be used eventually, not necessarily by those data providers.

And I'm just wondering, demand wise what you're seeing on your end in terms of the pace of construction of this stuff.

Well, the pace is insane.

I mean, everybody in the technology sector that is building the infrastructure.

You know, these companies that used to be selling software are becoming industrial companies they're building, you know, their Capex is more than some of the giant car manufacturing companies, so we think of them very differently.

But the, uh, demand is tremendous.

I think our business has doubled in the last year.

On the data centre side, there's two kinds of things going on.

Inference, which is the digitization of everything.

The, um, back and forth with your iPhone, your questions to Siri, that sort of thing.

That inference is booming.

You know, we we are hearing from some of our big clients 10 times growth in 6 to 7 years is, uh, kind of standard talk.

And, uh, that means that our business will double again next year on the compute side.

And maybe again after that, Uh, by 2020 26 A. I is, uh, the trainer model.

So this is not the computer that you interact with as much.

These are the learning buildings, and the demand for that is obviously off the charts.

It's taking up some of the inference.

So some of the utility that's already been built is being sucked up by by artificial intelligence.

But I think right now the gloves are off and everybody in the business is, uh is, uh, making massive infrastructure investments right now.

And, Bob, as you build out these data centres, um is activity um concentrated in certain states.

Certain regions more than others.

Yeah, entrances really kind of wants to be close to the use.

So the major cities, major metropolitan areas, um, pretty pretty spread out around population centres.

A. I is less dependent on that.

They need fibre.

And, you know, it used to be cheap power.

Now it's just power.

Where can I get power?

And even the land cost is not as much of an influence, but the A I trainer model buildings can can be almost anywhere.

Um, And when we hear about these things being built, you know, we we hear from the likes of NVIDIA.

They can hardly make chips fast enough to go into these things from your perspective and what you're building materials.

Are you seeing any kind of choke point?

Well, labour will be a limiting factor.

You know, some of these jobs have 800 to 1000 electricians on them, but I think all the technology companies are really focused on, um, using all of the building community resources today to start doing modular construction fab.

We're we're using the modelling like, uh, we've never used it before.

I think when you build a huge campus of data centres.

It gives you the opportunity to build a fab facility right next to the building and so we just have to find ways to counteract those limiting factors like labour.