Contact Us
News

Harrison Street's Data Center Head On Its New $600M Fund, DeepSeek And M&A

Fresh off the closing of a $600M data center fund, Harrison Street Managing Director and Head of Digital Assets Michael Hochanadel says the firm's focus on joint ventures across the broader digital infrastructure ecosystem leaves it well positioned for what lies ahead — even amid the uncertainty around the sector's artificial intelligence boom. 

Placeholder
Harrison Street Managing Director and Head of Digital Assets Michael Hochanadel

The firm announced last week it secured $600M in capital commitments for its first dedicated pool of investment capital exclusively for data center assets. According to Harrison Street, the fund will focus primarily on new development, with all capital deployed toward projects and assets in U.S. markets. The fund’s limited partner base consists of institutional investors like sovereign wealth funds, insurance companies and pension plans. 

While the fund is Harrison Street’s first dedicated exclusively to digital infrastructure, the firm is no newcomer to the sector. Since 2018, Harrison Street has invested more than $5.6B in a range of data centers and other digital infrastructure assets. The firm has made more than two dozen investments in the space, including more than 6.5M SF of data centers totaling more than 2.1 gigawatts of capacity. 

Its recent projects include data centers in Nevada, Virginia and Texas developed through a joint venture with American Real Estate partners-owned PowerHouse. Harrison Street also partnered with 1547 Critical Systems Realty to acquire a pair of connectivity-focused data centers in Indiana and Hawaii last year

Hochanadel has led Harrison Street’s data center and digital infrastructure push. He joined the firm in 2020 after holding senior leadership positions on JLL’s capital markets teams, with a focus on data centers.

Earlier this week, Hochanadel spoke with Bisnow about what differentiates Harrison Street’s approach to the data center market, the potential impact of Chinese AI startup DeepSeek, and what lies ahead on the data center M&A landscape. 

This interview has been edited for length and clarity. 

Bisnow: Harrison Street has been a pretty big player in the data center space for a while. You’ve deployed a ton of capital toward data center development and assets in the past couple of years. Why launch a data center-specific fund?  

Hochanadel: The basic answer to that question is really around the quantum of capital required for the sector and the size of each individual discrete deal in the sector. It's just very, very large transactions, particularly on some of these campus development projects. That’s a big portion of our overall strategy, especially when you think about it relative to the other types of things that we do at Harrison Street. One of the things that we've been increasingly doing is organizing capital dedicated to data centers alongside our flagship multisector funds as a way to balance the allocation exposure there.

Bisnow: Are you talking about funds like the $2.5B opportunistic fund you guys closed in September for investments into data centers but also student housing, self-storage and some other alternative asset classes?   

Hochanadel: Exactly. While data centers is definitely one of our highest-conviction sectors, there are a couple of other things that we focus on that are also in that high-conviction bucket — namely, student housing and senior living. Beyond those three sectors, there are a variety of other areas that we still do a bit more opportunistically based on the circumstances.

Bisnow: Can you give me a sense of how you plan to deploy the capital in the new data center fund? 

Hochanadel: Thematically, as a firm, we invest in a variety of different subsets of the data center asset class. One, obviously, is some of these large-format campus development projects — mostly powered shell and mostly in service of the hyperscale community. That's been a very large portion of what we've done to date and is a very substantial portion of our pipeline going forward. 

Beyond that, we invest in carrier hotels and colocation assets. It’s not the same dollar volume in those categories, just for the obvious reason that the development side right now is really where most of the new demand is going. But colocation investments, carrier hotels and then, importantly, development would generally be supported by the opportunity fund capital that we raise.

Bisnow: Along those lines, Harrison Street has had a pretty diversified approach in the past in terms of the different segments of the industry. You acquired a connectivity-focused data center in Hawaii last year, for example. It’s not all building hyperscale campuses. That’s a strategic choice? 

Hochanadel: We really like the way that the different parts of the ecosystem work together. You've got large campuses, generally with hyperscale customers, where they're operating significant scaled facilities with a couple hundred thousand square feet and 60 megawatts per building or somewhere thereabouts. Then you have networks of fiber connecting those facilities with each other and ultimately tethering back to a central point of interconnection.

That interconnection building is typically referred to as a “carrier hotel” and is the physical location where network providers interact with network end users. And then, as many of the operators across the markets have been moving upscale to accommodate hyperscalers, we've increasingly found that there's a cohort of end users that still need dedicated physical space within a colocation environment that have been overlooked by many market participants. That's an area we've continued to focus on strategically where we think it makes sense.

Bisnow: So, you guys see colocation as an underserved growth area in the short-term future?

Hochanadel: Exactly. It's the 1 MW or so to maybe 5 MW or 6 MW deployments. Increasingly, with all the focus on 30 MW-plus, in some contexts there's not even a willingness to do those smaller deals because there's a desire to keep contiguous space and power available in larger chunks. 

Many of those smaller requirements that are going into dedicated colocation space are not well suited to be served by the cloud for one reason or another. Certain financial services requirements, certain healthcare-related requirements and certain content distribution requirements often need their own dedicated infrastructure inside of a colocation facility.

Placeholder

Bisnow: So, besides this diversification and your convictions around colocation, what else distinguishes your strategy from other firms investing comparable capital in the data center space? 

Hochanadel: Being able to understand the end users’ needs across the various parts of their application stack and network ecosystem and having relationships with those different groups internally is definitely important. The information insights and advantages that come from working across those different parts of the overall business can be very helpful as we think about site selection and new markets.

Added on to that is the momentum and success we're having with our programmatic relationships with our operating partners. That positions us well to take advantage of the current market opportunity.

We have five operating partners across our digital strategy. One is explicitly focused on our fiber, Summit Infrastructure Group. The other four focus on the different categories of data center facilities. 1547 Critical Systems is our operating partner around the carrier hotel and colocation space. Then we have three groups that we work with on hyperscale development. There’s PowerHouse, which is owned by American Real Estate Partners — we have a very successful program with them that goes back five years. There’s Oppidan, which is another group that's very focused on that sector. And then Lincoln Property Co. is a group that we’ve executed in that bucket with as well. Those existing deep programmatic relationships we have with each of those groups are very useful for us.

Bisnow: I was going to ask about exactly that, the fact that seemingly all of Harrison Street’s big data center transactions have been joint ventures with a handful of operators. Is that just a matter of leveraging their expertise in different segments of the market? 

Hochanadel: It also allows us to structure our investments within the context of a conventional real estate joint venture. A lot of the infrastructure capital that you see flow into the sector flows in at the enterprise level, and it's not so cleanly oriented around specific underlying discrete projects. That's something that's worked well for our limited partners as well.

Bisnow: Pivoting our focus a little bit here, the big news this week has been about Chinese AI firm DeepSeek and its new model, which is supposedly far more efficient than what’s been developed previously. There’s a lot of talk about the implications for the data center industry, particularly the long-term demand picture, if it turns out that less computing is needed for the industry’s biggest tenants to train AI models than previously thought. What’s your view of the implications here, and is there reason for concern? 

Hochanadel: First, you have to caveat this whole conversation based on the credibility of the information that we're getting and understand that some of it may be true, some of it may be partially true or all of it may be propaganda.

To the extent that all of it is true and that there are implications on demand long-term, I’d argue that the implication there is narrowly focused on a specific subset of the broader data center market, and that’s these large campus locations that are explicitly focused on AI training. Honestly, we've intentionally shied away from those types of investments and those types of locations. It's important to us where we're developing or investing in data center assets is in locations that, while they could support AI training, they also could support other workloads and conventional use cases like typical cloud deployments, edge deployments or content distribution. Not just narrowly focused on AI training data. 

The locations, in our mind, that are specifically focused on AI training data are these ultratertiary locations where there's an abundance of space and power, but maybe not close proximity to network or end users. They're great for these large-batch processing applications. But long term, even before this DeepSeek announcement, there were some people saying that once the AI models are trained, those types of facilities would become superfluous and would have less defensible residual value. We already incorporate a lot of that thinking into our investment strategy.

The other argument you could make is that if some of these breakthroughs are real, it may actually give rise to a faster speed to market for a variety of the applications and use cases that will be built on top of AI. It'll be a sort of democratizing trend: more companies, more startups, more ideas being built on top of AI. The term that you'll hear that's been bandied about is “Jevons paradox,” the idea that as something becomes much more efficient, it proliferates and drives more demand than it was potentially destroying by virtue of that efficiency gain. You may see that play out. It would probably show up more on the inference side than on the training side. So, there's a strong argument to be made that there will be no impact on demand, or even an increasing impact on demand on the inference side.

Bisnow: So, if cheaper, more efficient AI training does create more demand from inference workloads, how does that translate into data center demand? What markets or segments of the industry will benefit?

Hochanadel: Broadly speaking, it will be the established markets across the spectrum of types of assets. Some of it's going to reside in cloud environments, and some of it's going to reside in edge deployments and carrier hotels. But it will be the places that are network-dense and where there's significant interaction with eyeballs. That's the key difference.

Bisnow: If more efficient AI training leads to even a slight reduction in data center demand over the long term, what does that mean for some of the massive investments that are really tied to the industry’s long-term trajectory? I’m thinking about the acquisitions of data center developers or energy providers and assets that we’ve seen. 

Hochanadel: In the near term, even if demand is flat year-over-year or declines a little bit, we're still adding 20% of the entire market each year. So, when you think about that stat in the context of any other real estate sector, you can't even comprehend what that means. 

A lot of the talk has been around power. Long term, we need to secure the generation of more electricity than we currently are planning. Full stop. That just needs to happen. It needs to happen for data centers but also for other things. Data centers historically have consumed a couple percentage points of the overall electricity generation. That's going to double or triple, but that's just high single digits or even low double digits in terms of percentage. In the coming years, we still need to generate more electricity for all the things that we want to do around electric cars, to support our manufacturing base. I don't think there are any long-term implications here relative to our need as a country to create more power generation. 

Bisnow: Finally, shifting gears again, 2024 was a record year for data center mergers and acquisitions. What do you expect to see in the M&A landscape in the year ahead?

Hochanadel: There’s probably going to be less conventional M&A than there has been historically. That's largely a product of most of the targets either having been acquired or aligned with capital partners. There will be a little bit here and there, but I don't think conventional M&A is going to be the major feature of the marketplace.

You're not going to see large asset managers coming in and outright acquiring platforms. The platforms are all married up, and the quantum of capital is getting to a point where it's just not possible.