Cool Vector Video-Podcast TRANSCRIPT: "How AVAIO Builds Data Centers with 'Just-in-Time' Capital"
Cool Vector Video-Podcast: Charting the rise of data centers and the digital infrastructure asset class.
Episode Title: How AVAIO Builds Data Centers with 'Just-in-Time' Capital
Episode Duration: 37:34
Originally Posted: March 10, 2025
Listen to the Full Episode here.
Speakers: Mark McComiskey, AVAIO Capital; MODERATOR: David Snow, Cool Vector
Episode Overview: Data center investment strategies that focus on ground-up development in unconventional locations are more complex, and therefore more likely to generate stronger returns, says Mark McComiskey, a partner at AVAIO Capital.
In an extensive conversation with Cool Vector, McComiskey explains that many private capital firms active in digital infrastructure invest in existing assets, or compete for sites in overheated hubs like Northern Virginia. By contrast, AVAIO is simultaneously developing six sites in less competitive markets, positioning itself as data center provider of choice to large customers.
AVAIO’s approach involves incremental risk management—deploying capital only as sites pass key milestones. This method ensures projects are fully de-risked before significant investment, reducing exposure to cost overruns or regulatory hurdles, he says.
At present, AVAIO is bringing to market a diverse portfolio of sites. “Instead of pitching one-off locations, we’re offering AI-focused campuses, cloud deployments, and hyperscaler-ready sites across North America and Europe,” says McComiskey.
The continued high demand for data center capacity is influencing negotiating dynamics between providers and customers. Customers in need of cloud and AI compute are willing to pay premiums for sites that can deliver in the next 24 to 36 months. “If you can build in 2025 or 2026, you have leverage,” says McComiskey. “If you’re offering capacity in 2030, the power shifts back to the customer.”
With billions pouring into AI-driven infrastructure, some market observers worry the sector is overheating. McComiskey acknowledges signs of speculation—like developers stockpiling electrical components without confirmed projects—but argues that irrational exuberance is still under control. “No one’s building speculative capacity without customers lined up,” he says. “Unlike real estate bubbles, where demand can disappear overnight, AI and cloud computing growth isn’t slowing down anytime soon.”
David Snow: Hello and welcome to Cool Vector. I'm David Snow, your host, and today we're joined by Mark McComiskey from AVAIO Capital. AVAIO Capital is a major investor in digital infrastructure and other forms of infrastructure. It's based in New York City, and its team has a lot of expertise in development, engineering, and construction.
So we're very fortunate to be speaking to Mark today. I'd love to get your thoughts on what's going on in the world of data center development and how AVAIO Capital fits into that ecosystem. You launched AVAIO Capital in 2019. It was originally a division of AECOM, which is a major engineering firm. It's now an independent private investment firm with, again, lots of resources and experience in construction and engineering. When you launched AVAIO Capital, what was your original thesis? Why did you believe that the world warranted the creation of yet another private investment firm?
Mark McComiskey: Yeah, well, that's always the question, right? We were birthed, as you noted, out of an infrastructure firm—an engineering firm—and had a focus on infrastructure. The strategy we used to justify the creation of a new investment firm in the space is that the infrastructure investment world is very similar in its fundraising approach.
They say, listen, there are trillions and trillions of dollars of infrastructure that need to get built. There's less money available than is needed. There's a funding gap. So you, Mr. Investor, should invest with us because there's demand for capital, and we can steer the capital to that demand.
But when you go and look at what those infrastructure firms do—and I worked for one—they buy existing infrastructure. They don't address the funding gap for the most part. There are a few exceptions, but the vast majority—90% plus—of infrastructure capital goes into existing assets. And that's where their expertise is: evaluating existing cash-flowing, operating assets.
But what that leaves unaddressed is the societal gap for capital and expertise to bring new infrastructure into existence. Our thesis was that if we focused on that—if we made that our niche—we'd be offering something the market needed.
Classical economics would tell you that where demand for capital is highest and supply is lowest, you get the possibility of better returns. And we'd have the advantage of not having a lot of competition because the vast majority of investment firms that focus on infrastructure were not all that interested in creating new assets, but would serve as a fantastic exit audience for the assets we were creating.
David Snow: You're currently backed by a $375 million investment from an investor. What did this investor like about your team that led them to put $375 million behind the strategy?
Mark McComiskey: We have a few verticals that we focus on. The three principal ones are digital infrastructure, energy transition—so renewables and things that help you get from carbon-intensive to carbon-free energy infrastructure—and transportation.
We have, as a team, a lot of experience in the energy side of things given our backgrounds. Everything from traditional power, thermal power, and renewable power to oil and gas, and infrastructure and the construction and equipment and services businesses that feed those sectors across North America and Europe.
We had a view that digital infrastructure was going to increasingly become dominated by data center investment, and that data centers were fundamentally going to become a type of infrastructure where power was a defining characteristic of success. So there was an overlap between our expertise in power in multiple regions and our strategy in the data center space that investors found interesting. That’s ultimately why they backed us.
David Snow: You've touched on this in your comments about energy being a necessary expertise when building data centers, but if you were asked what differentiates your team from the many other teams out there building data centers—or saying they build data centers—how would you describe those differences?
Mark McComiskey: I'll answer a slightly broader question, which is what differentiates our approach. The team is differentiated in the sense that it’s composed of a mix of people with long backgrounds as private equity investors who backed development teams in all kinds of things: renewable power development, traditional power development, development of transportation infrastructure, pipeline infrastructure—you name it.
So we have an investor perspective. We understand how money gets deployed, where it gets wasted, where the value creation milestones are in the development process. And we structured our approach to development in a way that minimizes capital deployment unless it is actually creating value step-up milestones.
But we also brought in—because in order to achieve those milestones and move the development along—you need to have people with experience in real estate, entitlement, real estate permitting, identification, and procurement. And as our sites have become more mature, we've added people with construction and project management expertise in the data center space. Our approach has been to start with a small team and work through the steps on a number of sites to get them to the point where we need incremental expertise and then expand the team.
It's just-in-time deployment of capital and just-in-time additions to the team to keep the cost base low but make sure the expertise matches the needs at the time. That’s a differentiation in our approach.
In terms of how the team is put together—you don't find a lot of people in the data center development space that mix all three. You'll find people who have worked in site acquisition for hyperscalers or procurement people for big data center operators. But finding a business led by investors with an investor mindset, land development and entitlement professionals, and data center construction professionals—that’s rare.
We wanted to move at least half a dozen sites along to the point where you could immediately start construction before we raised an additional round of capital and commercialized the sites.
The thesis being that a lot of people out there have a piece of land and say, “Hey, we’re data center developers.” But there aren’t a lot of businesses out there that say, “We can serve your needs in Europe and North America. We have sites that are good for AI. We have sites that are good for cloud. We have 100 megawatt sites and 500 megawatt sites.” We're coming out with a full portfolio of construction-ready sites all at once. And to the best of our knowledge, that's unique in this space.
David Snow: I know from the real estate asset class that when investors think about greenfield real estate—real estate that needs to be built from the ground up—there’s often a lot of talk around the risks of development. The project might not get completed, it might get delayed, resources might not be available—and of course that affects returns. Is that similar to how you think about risk management on the data center side? Is development itself a risk, and does your team have expertise in understanding and managing those risks?
Mark McComiskey: That’s exactly right. People use the word “development” to mean different things. For us, development means all of the activities that take place right up to the point where you start construction. That’s finding the sites, securing the power, getting a specific timeline on the power, getting the site permitted and entitled, getting site and construction plans approved.
Mark McComiskey: All of those things. When we've got all of those, then we can start construction. We have three phases in our terminology: development, which is everything up to construction; then the actual construction; and then operations. There is risk inherent in development, and that is what we have spent the time since 2019 working through.
Our investors felt that the value creation inherent in moving these sites to the point where they were construction-ready in the development phase offered a very attractive risk-return profile, given the way we do it. There are other groups, both in the data center space and in things like renewables, traditional power, or real estate, that take a much more heavy-handed, high-spend approach up front.
We dribble cash out as we successively de-risk sites so we don't have a lot exposed. We've frequently started down the path on sites and then stepped away because our assessment was that it was going to be difficult, time-consuming, or more expensive than anticipated to move those sites forward.
We start with a big funnel of projects and then winnow those down to the ones that are most attractive. We've ended up now with six projects that we're ready to put into construction in 2025. The ability to manage those risks—and in particular, the ability to constrain capital deployment until key risks have been mitigated—is something very important in development. It's been a huge focus of what we've done for the last five years.
David Snow: How would you describe your ability to source land—to source opportunities for these data centers? Is it your network? Is it your insights into different types of land and access to power and everything necessary for a data center?
Mark McComiskey: Yeah, that's a great question. The initial framework for how we went about looking for land was established back when we started this strategy in 2019 and 2020. That insight was driven by power. We looked at the data center industry at the time—we were very excited about it because of strong secular trends that looked like they would continue for quite some time.
What really struck us was that the industry was concentrating all its data centers in a handful of cities—or at least counties. Northern Virginia is kind of famous for it. Dallas is now the second-largest location. But there were a handful of cities in North America and Europe where everyone seemed to want to put their data centers.
We looked at how much power those data centers were consuming—this was before AI ramped everything up exponentially—and we said, within three or four years, those places are going to be power-constrained. Not necessarily in the ability to generate electrons, because there's a deep well of power generation in the U.S., but in the ability to get those electrons to the sites. The grid wasn't going to be able to handle it.
We weren’t going to pay a million dollars an acre for land and compete with everyone else in the five cities everyone wanted to be in. There’s no way this industry could continue to grow while staying focused in just that handful of cities.
So our view was to go to logical "step-out" geographies. We would find sites, work with landowners before the land was bid up, and procure power—because the grids in those places are different from the grids in the demand-concentrated cities. We wanted to have construction-ready land with significant amounts of power by the time congestion started hitting those epicenters.
We built a network—your suggestion about network is right. We identified about half a dozen geographies to focus on. Then we went out and built local networks to identify sites. We worked with local development authorities, landowners, and in some cases, real estate agents. We used databases to cross-reference all of that—looking at where high-voltage and medium-voltage transmission lines were, and doing a nodal analysis of which nodes in various ISOs would be less constrained.
Then we triangulated that with what the industry needs: data centers near reasonably significant urban populations where they can draw qualified labor. That was our approach to narrowing things down.
It’s worked well. If you think about Northern California, the epicenter is Santa Clara. Santa Clara now says they can’t deliver power until the mid-2030s if you go in with a fresh application. The town itself is becoming increasingly negative on new data center entrants. So even if you’re willing to wait until 2035, it’s not clear you could get permitted.
We’ve got a fully permitted site with 100 megawatts of power that’s just half an hour from there. Because we started three or four years ago, we were able to secure power and permitting in Northern California. We’ve done the same in Northern Virginia, and we’re doing the same in other geographies.
It was about building local networks in areas the industry wasn’t focused on when we started. That really was the essence of our sourcing strategy.
David Snow: Given that you are sourcing opportunities in areas not heavily trafficked by the data center industry—at least not yet—were you talking to landowners or brokers who had never done a data center deal before? Were they curious or surprised that you wanted to turn their land into a site for infrastructure?
Mark McComiskey: Yeah. At the time we started most of these sites, the answer is yes. We were dealing with people who hadn’t really heard much about data centers. Data centers weren’t talked about as much four years ago as they are now. These were people who hadn’t been approached to discuss data center developments and who were very interested, but didn’t have—and still mostly don’t have—the expertise to interact with utilities, pay for power studies, or even figure out what to ask for.
They were looking for partners. These were very sophisticated, successful landowners, but they were looking for people to help. That has worked well for us.
And the industry has moved in our direction. Our site in Northern California is now front and center for where people are looking to expand. Our site in Virginia, about an hour north of Richmond, was one of the early sites there, but now Richmond has become a hotbed and development is spreading up the corridor toward Northern Virginia.
We got into South Dallas early, and that market has moved in our direction. There are still places like Little Rock and Jackson that are earlier in that stage. That presents both opportunities and challenges.
That said, over the last four years—and especially with the advent of AI, which allows for more flexibility in site location compared to cloud service data centers—there is now a broader awareness of the data center market. Landowners are more educated today than they were four years ago.
Mark McComiskey: Sometimes creating unrealistic expectations and sometimes just making them savvier counterparties. It's kind of a mix.
David Snow: As someone with a lot of experience in engineering and construction, what are you seeing now? What do you predict for 2025 as far as the availability of construction crews? I keep hearing about electricians, plumbers—people absolutely essential to building out this infrastructure. Where do you see the pain points?
Mark McComiskey: The congestion that we forecast was grid congestion—the ability to flow electrons to areas. That has manifested. Getting new power into Santa Clara is a 10-year undertaking. There are parts of Northern Virginia where you can expect to wait five to seven years to get a significant new load connected.
There has, generally speaking, been a massive shortage of labor in the trades—engineering, architecture, electrical—and not just labor, but also certain key electrical components. High-voltage circuit breakers, for example, or generators—not the generators themselves, but ironically, the enclosures, the low-tech boxes that go around them. There have just been a series of bottlenecks. Not solely because of data centers, although data centers are a significant factor. Hundreds of billions of dollars are going into data center development right now.
But there has also been a widely publicized boom in industrial and manufacturing investment across the U.S. in the last two or three years—car factories, battery factories, chip fabs. You've got to put that in context. After World War II, the U.S. saw substantial electricity demand growth—at one point growing 1.5 times GDP. Then it slowed to match GDP. But from about 2004 to 2019, electricity demand growth was flat to negative. Utilities and manufacturers of power equipment got used to flat or shrinking markets, and they’re struggling now to ramp up production for what is now a growing market.
That’s further complicated by the significant renewable build-out, which requires all-new components, electricians, and construction staff. So if you're managing to a timeline in the development of data centers—or anything requiring technical construction and electrical equipment—you need to be very focused on supply chain.
It’s not just about finding a construction crew or an electrician—it’s about finding the qualified ones. The ones that have done it ten times and won’t make mistakes. So you can have confidence it’ll be done on schedule and on budget. You have to have relationships in the trades with the right firms and the right experience.
That’s often hyper-local. It’s not the same firm building for you in California that’s going to build for you in Little Rock or Virginia. It requires real industry experience. Same goes for procurement. You need a strategy around buying long-lead-time equipment in advance if you're going to commit to a firm delivery timeline for a customer.
David Snow: Is there any particular part or piece of equipment that makes you nervous? That makes you think, “Geez, I better order a bunch of these before the rest of the world buys them out?”
Mark McComiskey: I'm old enough to have gone through this cycle a few times. There are a few pain points—high-voltage circuit breakers, generators—but largely due to the generator enclosures, or medium-voltage switches. You’re seeing manufacturers quote long lead times.
What that’s done is caused everyone with capital to go and place orders—often without a specific project lined up—just on the belief they’ll eventually have one. The power industry has seen that behavior three or four times over the last 20 years. So have other industries. It means there are more orders on the books than there’s real demand for in the timeframe those orders will be delivered.
I'm not all that concerned, because we’ve got our orders placed and have specific projects to line them up against. But I expect that over the next 24 months, you’ll see those lead times shrink.
This shortage that’s causing panic right now will get better for three reasons: First, manufacturers are expanding capacity—Siemens and others are building new plants in the U.S. Second, many customers—utilities and data center developers—are qualifying new suppliers they didn’t use before. Some of these Korean and Japanese suppliers make equally good equipment and are now getting qualified because of wait times. Third, there’s been a lot of overordering, and eventually that elephant will move through the snake and come out the other side.
David Snow: For the six data center projects you've greenlit, have you identified customers yet for the servers?
Mark McComiskey: We're just getting out into the market to commercialize our projects. Our strategy was that we didn’t want to go talk to a customer with just one site. We didn’t want to say, “Hey, we’ve got this place in Virginia” or “this place in California.” We wanted to say: “We’re a big established player with funding behind us. We have multiple sites. We’ve got a lot of power. We’ve got the equipment. We’ve got our own design—although we’re happy to adapt it to your specific needs. And we can solve a lot of problems for you at once.”
David Snow: So that process culminated in your project development progress through Q4 2024, and 2025 will be the year you commercialize. What kinds of customers are you looking for? Are you focused mostly on hyperscalers, or is it a broader group?
Mark McComiskey: Hyperscalers can mean a lot of things. There's a universe of probably 20 customers at the top of our list. That includes hyperscalers—whether it’s for cloud or AI.
There are emerging AI service companies like CoreWeave that aren’t hyperscalers but are soaking up significant power and capacity to offer AI services. Then there are the big financial and commercial firms that still need space. Apple is still out there contracting for data centers. We’re negotiating with major financial firms that can take an entire 40 or 50 megawatt building on one of our campuses.
We have sites that range from 100 megawatts to 500 megawatts. At 500 megawatts, you’re looking at a hyperscaler or a large AI player who wants to grow a campus. At 100 megawatts, you can split it between two financial firms. So it really depends on the site and its capabilities—that determines the kind of customer we’ll go after.
David Snow: If you could characterize the dynamics around negotiations with potential customers, what would they be? Is there much more demand than there is supply? Who's got the pendulum of power right now?
Mark McComiskey: It is an undeniable fact that lease rates per megawatt are going up. That’s not due to anything we’ve done—we’re just getting started in commercialization—but if you look at the rate per megawatt-month or kilowatt-month for data center space in almost every market in North America and Europe, those numbers have been rising rapidly.
Some of that reflects cost—as shortages have emerged, prices have gone up—but it’s been rising even when accounting for the cost of creating a new data center. That suggests the pendulum of leverage, at least as expressed through economics, has shifted somewhat in favor of the data center provider, just given how much demand there is and how many potential customers are focused on near-term delivery.
It's becoming, in many cases, perceived as existential to have significant AI capacity or cloud capacity in 2025 or 2026. If you actually have capacity and have moved forward with development and started construction—and can deliver in that timeframe—customers are willing to pay up.
If you're in a different situation—where you say, “Hey, we have a piece of land, and if you give us a contract, we’ll finish the development and build it for you and it'll be ready in 2030”—you don’t have much leverage. The rates you can expect are lower because customers are savvy. They know they’re creating your business. They’re not constrained; they have more options on how to get capacity for 2030. They could start from scratch themselves.
So there’s a bifurcation in the market. Developers who have secure power and can deliver in the next 24 to 36 months—and who have the capital to move forward independently of a customer contract—have significantly more leverage. They have options about who to contract with. Developers that need a customer to commit before putting any dollars at risk have much less leverage.
David Snow: And therefore they have longer timelines, where customers get to dictate a far greater share of the economics. I realize this is a complicated question, but I’ll ask it anyway. About how long right now would it take to build a data center? I don’t know the right starting point—whether I’ve already acquired the land and have flat dirt, or it’s earlier than that—but if a customer says, “How long until this thing is up and running,” what’s the time range?
CLIP 5
Mark McComiskey: You get a lot of different answers to that, but if you narrowly define it—say all your permits are in place, you can start moving dirt tomorrow, assume it's virgin land with nothing prepared, and the utility will connect the power when you need it, and you have your design done, construction plans approved, and a decent construction crew—you could get to first energization of a data hall on a data center campus within 12 to 14 months.
Some people say they can do it faster. Some say 18 months. But 12 to 18 months is a safe range under those starting conditions. Then it’s just a question of how much money and how many resources you throw at the problem.
Amazon is building out a multi-building campus in Jackson. The pace at which they're going is extraordinary. They've clearly made it a priority, put in a lot of resources, and are going at it aggressively. It looks like they’ll build out the entire campus all at once—not just a single building—and from the start of dirt moving to first energization, it looks like it will take them 14 to 15 months. They do it as well as anyone, so that's a reasonable timeframe.
But it takes much longer between the time you identify a site and when you can start construction. The data center world isn’t immune from the broader U.S. development challenges. It’s slow to go through the process of getting studies done, securing permits, developing and submitting plans, and getting approvals at local, state, and sometimes federal levels.
That process is time-consuming. Even once you’ve identified and secured a site, you should assume it can take over two years in some jurisdictions. In other jurisdictions, it’s faster—there are places in Texas and Arkansas that are more friendly to development. If something is already zoned commercial or industrial, it’s much easier. But repurposing agricultural land in Virginia into data center-suitable land—that takes much longer.
David Snow: You mentioned Santa Clara as being a place where the community is perhaps less enamored of data centers than in the past, and that’s potentially a roadblock to development. How much of a factor is community pushback? I imagine it depends on the community, but have you seen a rise in apprehension?
Mark McComiskey: We see a full spectrum of reactions. There are communities that are pushing back. Santa Clara is well known in the industry for that. It also suffers from the fact that they don’t have any power left.
Mark McComiskey: And so even in the absence of community pushback, things would be slower there. But then there are communities that are like, “We would love to have somebody come build something like that here.” It will generate taxes for the community. It’ll generate construction jobs. It brings technical jobs that, for some of the communities these are going into, pay substantially more than the going rate.
Then you get some that are concerned. They're happy to have the development happen, but they want to make sure it doesn’t become a blight, doesn’t create too much traffic. A lot of it is about educating the communities you're going into. Data centers have far less traffic for a given footprint than a logistics center, and logistics centers have been sprouting up all over the country.
They’re better co-citizens, if you will, than logistics centers in a lot of ways. The reactions, as you say, are somewhat community-dependent. We as a firm try to assess the likelihood that the community will be supportive before we get involved.
We just shy away. We don’t want to be in a position of trying to force through a development in a community that doesn’t want it. We're happy to sit down with communities that have questions or concerns and address those. We have contributed or committed to contribute to expanding greenways, to helping renovate historical buildings that are near us, to giving back to the communities we’re in—normal and appropriate things as a developer of significant infrastructure in a location.
But we don’t want to force something into a community that, for whatever reason, has decided they'd rather not see that kind of development. That has tripped up a variety of projects in a variety of locales. You really have to assess whether what you want to do is compatible with what the communities you're looking at want to do.
David Snow: Just a couple more questions, Mark, and then I’ll let you go. Really appreciate your sharing your expertise with us today. I don’t know how many institutional investors you’ve been speaking to lately, but what is the right way to help institutional investors understand the potential return profile of digital infrastructure projects and data centers in particular? What's the right way to set expectations?
Mark McComiskey: The institutional investor space is enormous, and infrastructure has expanded massively. It has focused on digital infrastructure over the last seven years with an increasing degree of fervor and has become very sophisticated in this space. This is an asset class—data centers have been slotted into the infrastructure bucket, not the real estate bucket.
That asset class is very focused on looking at established assets. There are a few groups that have stepped into situations where you're ready to start construction and have a customer contract, and the only thing separating you from operations is 12 months of construction with a qualified firm. They can provide capital but will exact a significant premium for taking that construction risk.
CLIP 6
There’s very little capital focused on the development side—on moving projects to the point where they’re construction-ready. Most investors recognize that established existing data center assets are being priced at very low yields on cost—single digit, and often not even high single digit.
When you look at 25 to 35 times EBITDA being paid for data center platforms, the bet is that they'll reprice contracts as they roll off or build new data centers to average down cost. The capital that supports construction-stage projects looks for high single-digit to low double-digit returns but primarily takes on defined construction risk—with a customer contract, power secured, builder lined up, and construction contracts in place.
Returns on the development side are private equity-level returns. We're talking about 30%+ through that part of the lifecycle. You can choose to sell the project at the end of development and make multiples of your capital—four times or more—not just 1.5x or 2x. Or you can carry it into construction and enjoy a much lower cost basis and much higher yield on cost than someone who pays market price for a construction-ready site.
The question for institutional investors is: are returns in the existing asset market becoming so thin that they’ll be pushed down into the part of the lifecycle where more value is created relative to risk? That’s our thesis. So I hope so.
David Snow: Final question: how would we know if we were in a bubble of some sort—a data center bubble, perhaps driven by overexuberance around AI or something else? Are there signs you're looking for?
Mark McComiskey: There are some signs of a bubble. Let me start by saying I don’t think we’re in a bubble, and I don’t think the structure of the industry allows for the kind of implosion that the word “bubble” usually suggests. But that said, some behaviors are familiar. Ordering equipment without having specific projects, for example—that characterized the power generation bubble in the early 2000s when companies were buying gas turbines without knowing where they’d put them.
There’s been a massive run-up in land prices. Utilities are changing what they charge for basic things like a power study. All those signs tell you demand is very high, and you need to be careful not to be on the edge of a bubble.
For the data center sector to truly be in a bubble that could implode, you'd have to see people building spec—lots of square feet and megawatts available with no customers. That’s not happening. Some companies are moving forward with site preparation far enough to attract customers and show they can deliver in 12 months—that’s smart. But they’re not spending $5 billion to build something and then going to market.
No one’s really making money off AI right now—except NVIDIA. I have no doubt that the commercial proposition for AI will eventually materialize, but right now, a lot of money is chasing a profit model. So it’s possible we see consolidation or a slowdown in the next two to three years.
I’m not predicting that, but it's not out of the question. It wouldn’t be an implosion—it would just be a pause or recalibration. We’re not in a situation where there’s excess capacity, so I’m not too worried about a crash.
David Snow: Excellent. Well, Mark McComiskey of AVAIO Capital, thank you so much for sharing your expertise with CoolVector today. I hope we can have you back in the future to tell us what’s going on in the world of digital infrastructure.
Mark McComiskey: It was a real pleasure, David.
TRANSCRIPT END