The Challenge
The global surge in data center construction is largely driven by the explosive growth in artificial intelligence (AI), cloud computing, and digital infrastructure. According to McKinsey, nearly $7 trillion in capital expenditures for data center infrastructure is expected globally by 2030, with demand for capacity potentially tripling at a 22% CAGR over that period. This surge puts tremendous pressure on supply chains, including automation suppliers.
Why it Matters
Valves play an essential role in the cooling requirements for these high-performance data center servers. Reliability and efficiency are required to drive down operating costs throughout the facility’s lifecycle.
Three Takeaways for Data Center Project Success
- Importance of assessing the entire lifecycle in valve selection
- How to minimize thermal disruptions causing performance degradation
- How valves play an essential role in the operational integrity of cooling towers, chillers, air handling units, fire suppression systems, and backup generators.
Give my podcast with Emerson’s Austin Greaves a listen, and visit the Data Center Valves section on Emerson.com to learn more about how intelligent valve solutions enable data centers to stay resilient, agile, and online.
Transcript
Jim: Hi everyone, this is Jim Cahill with another Emerson Automation Experts podcast. It’s impossible not to hear about the continuous and amazing advancements in artificial intelligence. These AI innovations require ever-increasing computer resources. New data centers are being built globally and even being considered for beyond our globe into the reaches of space.
I’m joined today by Austin Greaves to discuss the role of valves in keeping the cooling systems that protect these data centers running reliably and efficiently. Welcome, Austin.
Austin: Jim, thanks for having me. Happy to be here.
Jim: Well, it’s great to have you share your expertise with our listeners today. I guess let’s start out. Could you please introduce yourself and provide some details about your background?
Austin: Happy to. So I’ve been with Emerson for about eight months or so. I joined in July of 2025. Before that, I was with a PVF [pipe, valves, fittings] distributor in Houston, Texas is where I’m based, where I was based with them and where I am based now. Before that, I studied petroleum engineering at Texas A&M. So it’s not necessarily fitting that a petroleum engineer has now transferred into data center valve sales, but I mean, I guess now you kind of mentioned it there, the horizon, I guess, is aerospace engineering and something to do with getting these facilities up in space.
But that’s been the short and skinny of my young career thus far. I’m really excited to be chatting about data centers, which again, eight months ago, it’s new to me, but now it seems like something that’s 24 hours a day what I’m thinking about. And yeah, happy to be here chatting about it.
Jim: Yeah, we all have interesting careers, from petroleum engineering to valve sales, me from electrical engineering to podcast host. So we all have interesting lives here.
Austin: Right.
Jim: I guess before we get into more of the details, can you paint a picture for us about what’s happening in the data center industry right now that makes this time we’re in feel so different?
Austin: The short answer is demand. Demand is something that never has been happening before. It’s certainly to a scale that is unprecedented. Back in 2015 timeframe, the cloud compute boom was occurring within the data center space and massive amounts of data centers being produced for really the same players that we’re witnessing here today.
The difference is, and you mentioned it, right, with AI, high performance computing and the like, there is such an immense demand of all the things that we’re doing right now, doing this podcast, all of the obviously AI-generated tools that are helping us be more efficient within work and within our personal lives for that matter too. And it takes a ton, a ton of server space to be able to accomplish this. Not only at the scale that we’re at right now, but at sort of the next level scale for some of these folks to monetize off of artificial intelligence and monetize in a very real and significant way that is going to be the next industrial revolution that we’re sitting in right now.
So that in terms of quantity and quantity, that looks like, I think, $415 billion market by 2030, which is I believe it a triple of where it’s at right now. So not only that, as we race with China and as there’s sort of these geopolitical influences in there as well, it just continues to accelerate this demand in this buildup that we’re seeing.
Jim: That’s incredible, a tripling of it in here. And that affects so many areas, both within our world of automation and even beyond into semiconductors and everything else required for this. So from your perspective, what makes data center projects fundamentally different from other types of projects that we do?
Austin: I’m going to try to kick off each sort of topic here with one word, and the word that I would use for the projects themselves is speed. Once again, an unprecedented speed to market that the developers and end users are trying to reach, and thus that trickle down effect happens to engineering firms, general contractors, mechanical contractors, and then us at Emerson as the technology and valve suppliers to these facilities. And speaking with engineering firms, I mean, they’re turning around designs within three months.
So from inception to online date for the first, let’s say, 100 megawatts to 200 megawatts of the facility, it’s happening within 12 to 18 months. I mean, that is something that’s not a familiarity to us when it comes to LNG or maybe the other oil and gas, chemical and refining markets that Emerson is so leveraged in. But I think we’re not the only ones witnessing that.
Like I said, the contractors, the engineering firms, our customers in the end are the ones that are feeling most of the strain. And the challenge is how do we meet them with those issues? We’ve seen and had conversations with some folks where because of this immense pace, they’re getting graded by developers and end users.
Contractors have basically competitions within each project or within each customer, like a Google or like a Meta, on how exactly they’re performing on a particular project to see, will they get on to the next one or do we kind of put them on the sideline? So equally as much, that’s the same challenge that we’re trying to overcome as we tertiary on that aspect are getting graded by some of these contractors.
Jim: That’s kind of the interesting thing here that I think they’re using AI to improve their speed and their performance in all of that in order to create more bandwidth for additional AI in here. So it just keeps feeding upon itself, I think. So when you talk about with project teams, what pain points do you hear most around specs and getting these jobs built?
Austin: With the time aspects, there is a ton of, I mean, there’s a time compression, right? So that’s resulting in specs being rushed. A lot of the times what we see is a recycled specification for valves, equipment for other, that have been applied to other facilities. And I think the realistic part of that is most of that can be contributed to standardization of these facilities. As they try to reach scale, obviously standardization is going to be critical in doing that.
But there’s sort of the other instance in that there’s this massive talent wave coming into the data center space at the engineering level, at the contractor level, at the end user level. And they’re kind of just doing the same old, same old, whatever has worked in the past certainly should work again. But that’s not always the case.
As densities increase, as we’re reaching more and more power demands, more and more heat output from the server racks, there requires a step change in the specification to make sure that not only is the facility itself keeping up with the current technology, but the components within it, i.e. the valves that go into the cooling systems, are also keeping up with those requirements and those industrial standards.
Jim: Now, I’ve heard terms, I’m by no means a data center expert, but I’ve heard terms like chips, racks, chillers when it comes to data center cooling systems. But where do valves really fit into the big picture on these systems?
Austin: I mentioned the $415 billion market that it is. And a lot of that is going to be in the chips, the CDUs, the computer room air handling units, all of the expensive, more capital-intensive equipment that goes into the data center is going to comprise most of that. Now there’s a really large chunk there that I kind of referenced to the CDUs, the CRAH units, the cooling system as a whole is really where the valves are most critical.
Additionally, in some places on the power generation side, fuel isolation and shutoff. But for the most part, when we’re talking about ensuring efficiency and power utilization effectiveness, water utilization effectiveness within a facility, we’re talking about those chilled water systems all the way from the central plant, maybe even extending back to the water treatment facility that could be outside of the proverbial fence.
And then all the way down to the server racks themselves. So especially if we’re talking about a facility that’s liquid cooled direct to the chip, then valves are entering very closely into those chips themselves. So very critical across the entire environment, but mainly within the scope of that chilled-water system.
Jim: And you see those chilled water systems in a lot of other industries. It’s interesting that you can have that industrial look around what’s traditionally you think of as this clean room with all the equipment in there, but I can see how that’s really critical. So with AI and higher-density racks proliferating, what does this shift mean for design and mechanical teams?
Austin: I kind of mentioned the cloud-compute boom back in those days. And I wasn’t really, I was in high school at that point. So I’m not sure this is just retroactive learning, but I think rack densities were maxing out at I think 100-kilowatt rack density was going to be really pushing it back in the day, maybe more average on the 10 to 20 kilowatt scale. which in today’s day and age of data centers is really nothing.
I mean, individual rack densities themselves are nearly equal or as big as entire data centers from that timeline or maybe from 20 years ago. So I kind of referenced it in the previous question, but I think the paradigm shift that we are seeing is this move from commercial specification, commercial equipment, commercial grade valves in particular to industrial grade valves, things that can handle, number one, the flow rate, all of the technical specs that are required for the system, but also can deliver even more efficiency to help reduce power utilization effectiveness [PUE], which is something else that I mentioned every single step that they can get down closer to that PUE of one, which is impossible, but more realistically getting from a PUE of two, which is really not performing well and costing a lot of money, not only to the end user, but causing a lot of social headache as well.
So getting that down more and more by using more industrial-grade equipment, products, being more efficient with the selection there has really been the biggest shift that I think we’ve witnessed over the past two years as this boom has accelerated.
Jim: Yeah, it just seems like with that huge increase in density and power consumed, the amount of heat being produced is just, and made it a much larger to a more industrial type challenge from the cooling standpoint.
Austin: I think back in the day, it was just kind of what butterfly valves can we throw into the facility? And at that point in time, all butterfly valves were relatively similar. But now it’s not only that, it’s what butterfly valve has by 2%, by 5% has a better C sub V rate. What butterfly valve is going to be most effective in installing going back-to-back to back, throwing in hundreds and hundreds of valves within this condensed timeline.
So it’s really caused the design teams, mechanical engineers, mechanical contractors to be extremely critical with the way that they look at not only the product itself, but where the product is coming from and how that can be most efficiently packaged, especially when we’re talking about automated valves. Don’t want to forget that piece. But yeah, absolutely a massive paradigm shift. That is a great challenge for us and it seems like it’s working out well thus far.
Jim: Those are excellent points there. So for a large data center campus, what kind of valve scope are we actually talking about?
Austin: Redundancy is key and not even redundancy within multiple chillers and multiple pumps and multiple power generation points. But what we see is a ton of redundancy just within the individual piping system. So as we’re essentially the focal arm in protecting not only the expensive equipment, but the expensive information that sits within those chips, the proprietary information that sits within those chips. There’s a ton of redundancy. I mean, valves almost stacked on top of valves in some cases where in other industries and situations, you would think that seems like a lot.
However, in this instance, it’s really not because the second a CyrusOne facility goes down, it’s not only are emails not sending, it’s an American Airlines flight that can’t take off. It’s massive infrastructure problems that they can’t afford. And definitely as the valve supplier, we’re going to do our best to provide them robust products that eliminate that.
Now, the scale aspect is that it’s robust products that 2000 valves per facility. You know, when we think about hospitals or other HVAC applications where there’s a similar chiller set up with service piping and things like that, it could be 100 to 200 valves. So really we’re talking about a 10x, sometimes a 20x, depending on the facility type, the redundancy measures, the tier of the facility. It really is 10x is, I guess, what I could say to quantify the scale in terms of typical HVAC commercial application to what these data centers are demanding.
Jim: Yeah, that redundancy is critical. If you’re not maintaining the proper cooling, putting at risk all that compute power there is, it’s understandable why they want redundancy on top of redundancy to make sure things keep going. If there’s a problem with one particular area, you want to be able to work around that and keep going there. So when you review typical specs, where do you see the biggest gaps or missed opportunities?
Austin: I kind of hit on it a little bit with sort of the talent pool that’s coming in. You know, engineers from other industries are coming in to this space and driving a lot of the specification using what they know and what they know is typically really quality information.
But I think what we see as we go from facility to facility is there’s a massive delta in the consistency between project to project or application to application, what actually is required versus what actually is needed. And additionally, kind of converse to that, there’s some complacency in the designs themselves and that it’s kind of just we’ve used this before, so why not use it again?
Now, however, I kind of mentioned that the goal is ultimately for the end users to step down their PUE from whatever it currently sits at to getting as close to one as possible without the expense being downtime. And it seems like as we can step in and become a partner, there’s definitely moves that we can make to help progress that proper trend down to lower and lower PUE.
Again, I think I mentioned it through lower Cv, through more efficient flow and more efficient sizing, not only of the valve itself, but the actuator package, things like that. Yeah, I guess that would be probably the biggest missed opportunity is either complacency or just general driving the same ship that they’ve always used because we’re trying to reach scale, we’re trying to reach something at the expense of the other.
Jim: Yeah, just given the huge power consumption, anything you can do on the efficiency front, pays out over the life cycle, I would guess. So, I mean, just to paint a picture for it, so what happens in these plants if there’s a valve failure?
Austin: Number one, it could just cause uneven cooling. the second that there’s uneven pressure, uneven temperature, let’s say a valve is having leak by or a valve isn’t optimized to a proper flow rate or it’s leaking out maybe the stem journal or something crazy like that. Yeah, we’re going to have some serious power utilization effectiveness problems. We’re going to have some serious uneven cooling. which, they have stop gaps in place for all of these things, right?
But, the second that there’s uneven cooling within the facility, some operations person is likely going to know that that’s occurring. However, what I would say is, they, and they build in redundancy exactly for that circumstance. But the second that one issue occurs, one failure occurs, you have leak by, you have an improper seal that’s now damaging a pump or beginning to damage maybe some other expensive part of the system, if not the entire system itself, you’re opening up for another problem to cascade and then another problem to cascade. And all of a sudden, all these redundancy measures that have been put in place are all at risk because there’s been an improper valve selection.
There’s been a less robust valve put into the system that probably should have warranted a little bit different engineering or different manufacturer selection. So it really, on the surface, I would say it seems like it’s not a big deal. We have redundancy built in, but. What happens when all those things trickle and they all start to happen at the same time? And then all of a sudden operations is dealing with something else on the power side. Now we’re not as in tune with what’s going on in the cooling system there. Leak happens, gets onto some of the electrical equipment, starts a fire, and now worst case scenario is upon our hands.
So I mean, that’s kind of blowing it out of proportion, I think. But very realistically, when the goal and the ultimate KPI for the facility owners is, I mean, minutes of downtime per year, that’s really critical. And it’s something that shouldn’t be overlooked as we, again, move quickly through these projects and try to reach scale.
Jim: I think that point you made about, it cascading and causing other things to pumps or elsewhere in the system is really what you want to avoid. And with that redundancy, it lets you go take care of something right then and there, not wait, like a refinery for the next turnaround five years later. So yeah, that’s good from that standpoint. So how does valve selection enable a competitive advantage? I know you’ve touched on different areas of this, but what’s its role?
Austin: I kind of touched on it with the social aspects being the biggest extrapolation that I could think of right now. The power utilization effectiveness, the water utilization effectiveness, the big hot button items that drive these facilities, not only from their own KPIs, but also socially, how they get produced and erected.
For valve selection in particular, as it boils down to it is, I mean, optimization is key. And every single lever that the developers can pull to get their system more optimized you look no further than the valves themselves. getting to a lower Cv, improving flow performance, improving robust measures that are ultimately going to save the facility. Again, we kind of just talked about it.
When the worst happens you can be assured that I put that valve in there for a long time and I’m going to sit it and forget it because I’ve selected a robust valve that’s going to be effective for the system and it’s not going to compromise my expensive capital equipment. So I think there’s a massive competitive advantage there when it comes to partnering with engineering firms, with mechanical contractors, all the way up to developers for the most efficient solution when it comes to butterfly valves, check valves, ultimately the guardrails, the protection of the system itself.
Jim: Yeah, I mean, it all ties back to the efficiency of how you’re using that power that’s feeding everything. So it’s just critical there. So I mean, even if there’s a perfect design on all this, what role does the supply chain, just given the sheer volume of all the work being done here?
Austin: This is the supply chain aspect is the biggest opportunity although it’s a challenge that we see. And by that, I mean a challenge for the developers, for the contractors. We talked about the schedules that 12 to 18 months to get what some facilities might demand is 2,000 to 3,000, even 4,000 valves in one individual building, let alone the goal is to produce these buildings in parallel sometimes.
So talking about a one-gigawatt facility We’re 10, 100-megawatt facilities that the second that the first one is erected and we have walls, now we’re ready to start doing the mechanical components. Let’s get going on the second one. And we’re moving up months and months of lead time to accomplish this task, again, within 12 to 18 months. So the challenge for valve supply chain is immense.
Again, and it goes into the density increases, as we are producing more heat from the servers themselves, we need more cooling power to cool them down, i.e. we need much larger valves and we need a lot of them. Not only now at the cooling towers, all the way throughout the system, we need really large valves and we need a ton of them, which are for a 20-inch valve, it’s two to a crate that we’re shipping across the Pacific, let alone they need 500 of them.
So we’re talking about 250 crates, football fields of equipment that they’re asking for in 8 to 12 weeks, realistically. So it’s a logistics challenge. It’s a staging challenge. It’s a massive issue that we have to address that our customers, we support them in addressing because ultimately the end user knows that they’ll be able to get it from somewhere for somebody that wants to be on the project bad enough. So yeah, 2000 valves, 8 to 12 weeks, getting it there and being efficient with our delivery. is absolutely crucial.
Jim: Yeah, that’s an immense challenge. Just visualizing those football fields full of these valves. Let’s wind things down here. If you were giving the industry a checklist, what would that checklist include?
Austin: I think on the supply chain, no number one, I guess a partner, if I’m talking to a developer, a partner that like Emerson, obviously, has global strength to ride through the waves of demand spikes, the waves of change in technology. I mean, months ago, we hadn’t seen the scale of large-diameter valves that we’re seeing now. But now that we’re perceiving that, we have the ability to really ramp up and stay there, whether that wave rides back down or not. So I think that’s number one, just a large breadth of not only product, but scale of supply chain.
Proven experience within data centers and not only just the install base, but working through project execution is the other bullet point. Again, it goes back into the supply chain scenario, but knowing exactly how this project is going to run from building one all the way to building 10 or however many it’s going to be.
Long-term reliability and support. Kind of mentioned this during the conversation as well, that the valves are intended to be a set it and forget it. And those are the products that we produce and deliver to the market. So it’s not only that, but when things do go wrong, when the worst comes to worse or there’s some checklist that needs to be done to ensure, hey, the valve in this system is good, we’re having a problem elsewhere, life cycle support for all of these facilities. So those two sort of go in tandem, the reliability and support aspects.
Overall, I think just that at the scale, at the pace, It’s really easy to say, we just need to go to some consolidator. We just need to go to someone that can buy it all and distribute it all. And it doesn’t matter if they’re the manufacturer or not. No, the real differentiator here is having a vendor that’s not only just a vendor, they’re a partner. They have the manufacturing strength, they have the service strength. full ship of men and women to be able to tackle the immense challenge that happens from the engineering side of things all the way through to project execution.
Jim: I think it is about considering that entire life cycle, not just the project of doing it, but ongoing with what’s required through the life of those data centers and having someone, as you said, not just really a vendor to sell you some stuff, but more of a strategic partner to work with over that life cycle.
Well, thank you very much, Austin. This was a great conversation. And everyone listening here, you can check out the data center valve section on our Emerson.com site, and I’ll provide a link to it in the transcript or just ask your search or your AI buddy of choice for Emerson data center valves, and it’ll take you right to the information we have.
So thanks again for sharing your expertise, Austin, with our listeners.
Austin: Thanks, Jim. Happy to. Appreciate it.
-End of transcript-
