Real Estate Fund

Market Overview:
It’s All About Supply

Nothing dulls the senses more than a one-dimensional narrative. That is exactly what has happened in the real estate sector, and it’s been going on for almost two years. Sure, other issues have grabbed the spotlight from time to time, like return-to-office (RTO), for example. But no single factor has mattered more to publicly traded real estate securities than the path of interest rates. It seems anyone with even a modest connection to the markets is being incessantly polled for their thoughts about the most likely glide path of interest rates. It has become a circus of oversteering. Of course we’ve always believed difference of opinion makes a market, and this time is no different. However, as the market dwells on interest rates, we are taking this opportunity to look deeper for long-term value.

For management of Davis Real Estate Fund we believe the evolving supply picture is the opportunity. This is something real estate pundits watch closely, but it has garnered more attention over the past couple of years as demand has softened, even for charmed sectors like industrial. For a while now readers have been peppered with stories of surplus retail, too many apartments, and a sea of underutilized office space, among similar tales. And all of it is mostly true at the moment. Where we diverge from the pack is in our willingness to consider the long-term supply picture. On that basis we believe commercial real estate is in the early stages of supply constraints that will lead to a robust recovery in fundamentals.

For far too long money was simply too cheap and too plentiful. It allowed even the riskiest of developments to break ground. Even the pandemic failed to curtail development as the Federal Reserve was forced to keep rates low and the U.S. government allowed the money supply to balloon in an effort to sustain the economy. Whether that was the correct decision or not is moot because the supply of property in most sectors has now overwhelmed demand, leading to softening rents.

The promising outcome is that the Fed has had to increase rates materially over the past two years to curb inflation, leading to a virtual halt in construction lending. Banks and lenders of all stripes, struggling to deal with credit issues in their existing loan books, have essentially exited the market. It means that the only development occurring at the moment is for projects that were in the pipeline years ago and had funding commitments which predate the Fed’s rate hikes. New development starts have dropped precipitously. Given the long-dated nature of real estate development, the foundation for much improved property fundamentals is being cast.

Two Sectors of Interest

We can generalize by saying supply is slowing for all property sectors, but there are two sectors in particular that interest us. First is the apartment sector. Much has been made about the massive increase in apartment units slated for delivery this year and next. That is particularly true for Sunbelt markets where developers sought to get in front of strong migration trends. A germinating supply imbalance is something we saw coming in early 2023, but thus far demand has been sufficient to absorb newly delivered units without too great a perturbation in prices. It is an unexpected outcome and as a result the apartment sector has performed very well this year, although leadership in the sector is varied. We’ll get to that a bit more later on.

Of course, the sector must still deal with ongoing supply pressure that we expect will extend well into 2025. In the end, this pressure may cause the apartment sector to surrender gains, especially if a recession does take hold. However, looking further out suggests far better times. Over 400,000 units are expected to be delivered in 2024, some of which will surely get pushed into 2025. Deliveries in 2025 are expected to drop to roughly 200,000 units, excluding any amounts deferred from 2024. However, even pushing some units into 2025 means deliveries are likely to slow meaningfully. Given typical lease up times, it’s our belief that the excess supply of apartments will be absorbed during 2025. Pricing may soften to get those units occupied, but long-term we see things swinging back in landlords’ favor, potentially as early as the middle of next year in some markets.

“Said simply, the best office buildings are filling up and everything else is emptying. To boot, pricing for premier assets is not only resilient, but increasing in many cases.”

However hyperbolic this may seem, office is the other sector we believe is set to benefit from a favorable supply picture. You would be justified in asking, “How is that possible?” The answer requires a willingness to accept that not all offices are created equal. We have been on record with many of you stating our belief that demand for the best office space would win at the expense of poorly located commodity office product. That is exactly what is happening as employers use the benefits of lifestyle office, replete with amenities you’d likely find in a hotel, to drive employee desire to be in the office.

Over the past two years we’ve watched demand crowd into these buildings, and even though incremental leasing often represents a downsizing of prior space, share gains for the best buildings are offsetting this. According to Jones Lang LaSalle, since the onset of COVID the sort of buildings we are referring to absorbed 125 million square feet of demand. All other buildings lost 386 million square feet of occupancy. Said simply, the best buildings are filling up and everything else is emptying. To boot, pricing for premier assets is not only resilient, but increasing in many cases. This migration, which admittedly has been slow in coming, is setting the stage for an interesting recovery for some of the fund’s top office holdings like Cousins Properties and BXP, Inc. (formerly Boston Properties).

The logical question to ask is what happens if demand does indeed overwhelm the available supply of apartments and premier office space. Conventional wisdom would say build new product to satisfy that unmet demand. In principle that is the correct answer, but for the economics of development to work, market rents need to be at a level that justifies new construction. In the apartment sector, that is generally true even if you assume initial yields are creeping up to accommodate higher capital costs. For the office sector, however, market rents are way too low to justify new construction. That is particularly true given that initial yields in that sector not only need to exceed higher capital costs but also provide compensation for the risky veneer that office has acquired over the past few years. What all this means is that new development has slowed to such an extent that even if lending markets return to support new starts, we are looking at several years where there will be too little supply. That is not a bad environment in which to own commercial property, even office.

Arms Race for Data Space

We would be remiss if we didn’t acknowledge that the evolving supply picture looks favorable for all commercial real estate sectors. Thanks to a stagnant lending environment there is reason to believe the next few years will likely be a story of too few options for anyone seeking space of any sort, even if a recession means demand is otherwise weaker than expected. Yet there is one area where we do not expect weakness: data centers. Artificial intelligence (AI), broadly defined, has sparked a metaphorical arms race as technology companies of all stripes vie for dwindling availability of data center space. To boot, most believe we are only at the beginning of a decade-long push into AI that will hugely benefit incumbent data center companies like Digital Realty and Equinix.

We need to be mindful of history, however. Too often in the past, fundamentals in the data center sector failed to meet expectations. While there have been periods of considerable demand, new development always seemed at the ready. That “build it and they will come” paradigm placed pressure on renewing rent levels. As long as we’ve been covering data center companies we’ve marveled at the yields earned on new development and lamented the lack of same-store growth as renewal rental rates declined. Generally speaking, that’s why fund management has shied away from getting too involved with the sector.

Now, however, the meteoric rise of AI has changed the playing field considerably. The existing inventory of data center space is insufficient to meet current and projected demand. Currently there are about 35 gigawatts (GW) of data center inventory across the globe. (Sizing a data center has become all about power. Conventional units of area, such as square feet, are not very useful.) That includes primary and secondary markets as well as data centers owned by hyperscalers such as Meta and Amazon. Total demand is forecast to grow to 100GW by 2030. Given that current utilization of space in operation is 95%, and property under construction will add an incremental 7GW by the end of 2024, the near totality of which is spoken for, the implied demand growth is 18% per year through 2030. All other things being equal, that would make for heady days for anyone in the business of developing data centers.

“Those with a pressing need for data center space are going wherever they can find power. Underutilized substations in any state with below average power costs have become targets for anyone looking to deploy certain AI workloads.“

Unfortunately, the situation is not that simple. The immutable truth of AI is the more we use it, the more power we consume. To make matters more complex, AI workloads consume roughly ten times more power than conventional ones. Therein lies the chokepoint for data center developers. There simply isn’t enough power. To hear it from Ben Jenkins, CEO of DigitalBridge, partial owner of Vantage Data Centers, the power necessary to deploy all the chips NVIDIA plans to sell this year simply doesn’t exist, much less the power needed for the chips they, among others, plan to sell in the years to come.

Based on our research, we don’t believe the situation is being exaggerated. Moratoriums on new power authorizations are already in place in some markets, most notably Loudon county in northern Virginia, the largest data center market in the world. The existing power transmission infrastructure there is insufficient to meet developer demands for new data center starts. Until that is remedied—by late 2025, it’s hoped—no new developments will be allowed to press forward. Similar issues with lack of power availability or inadequate transmission are plaguing many of the primary and secondary data center markets in the world.

As a result, those with a pressing need for data center space for latency agnostic workloads are going wherever they can find power. That has pushed development away from incumbent markets like northern Virginia and Silicon Valley into places like Wyoming and Illinois. It seems that underutilized substations in any state with below-average power costs have become targets for anyone looking to deploy certain AI workloads.

This presents an interesting situation for the public data center REITs. Demand for data center space is as high as we’ve ever seen it, leading to a rapid recovery in renewal rents, but they can’t ramp development quickly enough to meet rapidly growing demand. Further, neither Digital Realty nor Equinix, the two public data center REITs in Davis Real Estate Fund, have material operations in markets like Wyoming and Illinois. Rather, they are concentrated in primary markets where power constraints are most likely. We hear from both companies that they want to ramp development as quickly as possible, but they freely acknowledge there is only so much they can do. Fortunately it does present a unique benefit for their existing property. As tenants renew, each company can maximize rental levels, given a dearth of competing space. Even absent new development deliveries, strong core growth should lead to outsized earnings growth.

When considering Digital’s and Equinix’s valuations, it is important to take into account the favorable operating conditions and structural forces afoot in AI. Both stocks skew fully valued in our models and have for some time. The question you should ask is, if they screen expensive, why are they both top 10 holdings? The answer is this. We know AI is in its infancy, it remains to be seen how steep the demand curve becomes, and we might be wrong about our growth rate assumptions, but an upward sloping curve is all but certain, and because of that we are being a bit more tolerant of higher than normal valuations.

Performance and Positioning:
Survive and Thrive

Performance during the first half of 2024 lagged our benchmark, with leadership coming from a few areas. As a group the fund’s apartment holdings were the standout performers year to date. As we mentioned earlier, that is despite a deluge of apartment deliveries so far this year and expectation that levels will remain elevated through 2025.

What has surprised investors is how resolute rents remained in the face of new supply. It’s a classic case of things being less bad than expected. Our position in the sector has worked well so far, particularly for coastal apartment REIT Essex Property and AvalonBay Communities. Lately we’ve been reducing the fund’s exposure to both given that their valuation discounts have become thinner. Proceeds have been moved into Sunbelt apartment REIT Camden Property Trust, where supply concerns keep valuation discounts wide, and into other value opportunities outside the apartment sector.

Digital Realty has also been a strong performer this year. After many years spent fighting sluggish rent growth and higher than average debt levels, Digital has really turned the corner. Much improved market rents have buttressed performance and several well-executed capital events have reduced the company’s net debt/EBITDA ratio to 5.4x (as of 3/31/2024) from 7.1x at the beginning of last year. As we mentioned, Digital’s valuation reflects those things, but we are holding firm on the belief that long-term growth rates are biased higher.

We’ve also looked for other ways to capitalize on AI trends. The business case for AI training models is well-documented, but we are also keen to look farther down the road to potential use applications. Inferencing models, or what we like to think of as the things that will make AI useful, will most likely have low-latency requirements. Some estimates suggest the total addressable market (TAM) for inferencing models could be 15 times greater than that for large language learning models.

"With American Tower and Crown Castle, we believe the worn-out bond-proxy narrative is set to exhaust itself as interest rates stabilize (or drop) and investors begin to better understand the scale of inferencing’s TAM.”

Inferencing workloads will be located in data centers adjacent to connection infrastructure that allows for the near-instantaneous transfer of information. In many instances those applications will require wireless infrastructure of the sort owned by American Tower and Crown Castle. The fund’s ownership of those two stocks over the past couple of years has hampered performance. (Despite being two of the largest public REITs in existence, neither are part of the fund’s benchmark.) It seems the market is quick to view both as bond proxies and, as interest rates have gone up, stock performance has suffered. We believe this worn-out narrative is set to exhaust itself as interest rates stabilize (or drop) and investors begin to better understand the scale of inferencing’s TAM.

In an echo of several recent shareholder letters, the one area where we have yet to have our value stance vindicated is office. Even though data has consistently supported our stance that the best lifestyle office properties will win at the expense of commodity product, our office holdings continue to remain moribund. Cousins Properties and BXP continue to trade more or less inversely with interest rates and struggle to gain investor attention. Discounted valuations, strong leasing and an ability to refinance maturing debt on attractive terms hasn’t been able to lift prices meaningfully. Nevertheless, results matter and we expect that deep discounts will eventually win investor interest.

As we head into the back half of 2024 and beyond we hold fast that the fund is primed for better days. Many of the forces that have worked against not only Davis Real Estate Fund in particular, but real estate in general, are loosening. We maintain a strong weighting in lower-risk real estate companies that we’ve often characterized as “forever investments”, with just enough exposure to deep value that over time should produce outsized returns. We are not trying to minimize the threats facing commercial real estate. There will be plenty for the newspapers to write about over the next couple of years. What we are saying is the fund is positioned in the companies we think will not only survive long-term, but also thrive.

In conclusion, as stewards of our clients’ savings our most important job is growing the value of the funds entrusted to us. With more than $2 billion of our own money invested alongside that of our clients, we are on this journey together.¹  This alignment with our clients is uncommon in our industry; our conviction in our portfolio of carefully selected companies is more than just words. As always, we appreciate your confidence in fund management.