Digital Realty has transformed its business from one that merely provided large companies vast amounts of space and power (a typical wholesale data center) to one that can offer customers of all sizes the full spectrum of space, power, and connection needs. Prior to its acquisition of Telx in 2015, Digital Realty garnered almost 95% of its total revenue from wholesale data centers and had no interconnection revenue. At the end of 2018, wholesale space comprised 77% of revenue, while co-location space provided 13% and connections generated 10%. We think the trend toward co-location and the need for interconnection will continue as enterprises relying on hybrid cloud models need to connect with multiple providers and data use rises. In our view, Digital Realty was smart to get into the more attractive co-location and interconnection business, and we think its ability to provide those services in conjunction with the capacity to offer wholesale space to the largest cloud providers leaves it well positioned to win in an evolving technological landscape, where huge cloud providers drive the industry but need to connect to virtually all other enterprises.
We think interconnection is the key that makes co-located data centers much more valuable than simple leasable real estate. In Digital Realty's data centers, tenants can directly connect with each other, resulting in reduced latency and superior security. Even when in different locations, Digital Realty customers can bypass the public Internet to connect with other Digital Realty data centers via direct fiber connections within cities or through a software-defined network between cities. With many networks connecting to data centers to move traffic, data centers are also prime locations to connect to the public Internet and for networks to peer with each other. We expect connections to become even more important as enterprises increasingly rely on the cloud for applications they traditionally performed in-house, and we believe the Internet of Things, artificial intelligence, and augmented reality will contribute to data center demand, all driving Digital Realty's business.
We assign Digital Realty a narrow economic moat, the result of cost advantages and switching costs. We estimate Digital has earned adjusted returns on invested capital, or ROICs, of just over 8% annually, on average, since 2009, narrowly exceeding the firm's 8% weighted average cost of capital, or WACC. We expect returns to steadily improve throughout our forecast due to the company's shift toward the higher-margin co-location and interconnection revenue streams.
Digital Realty, has traditionally been a wholesale data center provider, meaning that it has tended to lease very large amounts of space and power to relatively few tenants per facility. It gets almost 80% of its revenue (and at least 18 of its top 20 customers) from cloud, information technology, network, and content companies, with the remaining coming from financial services (12%) and other enterprises (9%). Digital's business model has been in contrast with rival Equinix, which leases space by the cabinet to tenants that use the space primarily for networking functions rather than storage and computing needs. However, following Digital's Telx acquisition in 2015, it has placed greater emphasis on co-location and interconnection, which typically lead to better returns.
In our view, switching costs provide Digital Realty's biggest competitive advantage, as there are numerous reasons why tenants are unlikely to churn to a competitor, namely the significant monetary costs to deploy equipment and the risk in moving critical equipment. Digital Realty, citing statistics from Align Communications, estimates that a client deploying equipment to use 1.125 megawatts of power—a fairly large customer, but the type Digital Realty traditionally caters to—would spend $15-$30 million to set up its equipment. Migration of equipment from one facility to another would cost an additional $10 million to $20 million and bring with it a host of other challenges. Migrating to a new data center often requires the services of a company like Align, which specializes in it, and typically includes extensive planning, a multistep process, and involvement of many people within an organization. Migration risks include down time, which can be extremely harmful and expensive. Migrations thus typically include equipment mirroring, which is costly, to guard against down time. Given the monetary costs and business risks, we don't expect that tenants take a decision to move lightly, and we think they are effectively precluded from price-shopping once they are in a data center. Digital Realty claims most churn results from tenants merging or no longer needing a data center presence because of company-specific issues, such as their own business struggles.
We think Digital's cost advantages stem from its scale and property ownership. According to 451 Research, the two biggest data center providers in the world, as measured by annual data center revenue, are Equinix and Digital Realty, which collectively accounted for 15% of the market as of 2016. Both companies have around 200 data centers across at least four continents, but Digital Realty has about 60% more leasable square feet than Equinix, and its operating margins have averaged 25% since 2010 as compared with 19% for Equinix. Given that two of data centers' biggest expenses are cooling equipment and power, we think scale allows a meaningful cost advantage, as a firm of size can contract for these needs in bulk. Digital Realty has said it is one of the largest consumers of generators and air conditioning equipment in the world, and it receives modest discounts on them as a result. It also has recently begun negotiating power needs on an aggregated basis within given metropolitan areas rather than for each campus individually, leading to even more favorable deals with electric utility companies. For example, the company recently locked in rates in Chicago through 2022 at a price 20% below the current market and secured lower than market rates in Texas.
We are convinced that the current environment results in a sustainable competitive advantage that makes it likely Digital Realty will earn excess returns over the next decade, but the evolving nature of technology keeps us from labeling Digital Realty a wide-moat firm. On balance, the advent of the cloud has been good for data center businesses, as the number of firms reducing their own data center needs has been much more than offset by those that increasingly move off-site for interconnection with cloud service providers for their hybrid cloud needs—the most common model for large enterprises. Data center companies are excited about the next wave of the cloud, with Digital Realty indicating that it stands to benefit greatly from artificial intelligence, the Internet of things, autonomous vehicles, and other technologies that require exponentially more data. However, the speed with which technology changes and data centers' close connection to it leave us wary about making a 20-year assumption with any confidence.
In addition to our concern that enterprises will have less need for data centers as data storage and computing power is concentrated in a few cloud service providers, hardware advances and software virtualization continue to increase capacities for a given physical size. As a result, customers that remain data center tenants may need less space. Also, as cloud service providers become even more dominant, we expect they will have increasing control over data centers' fortunes. While they seem to be great customers now, we can't bank on that holding true in the future.
We are maintaining our $112 per share fair value estimate for Digital Realty Trust, which implies an EV/adjusted EBITDA multiple of 19 and price/AFFO multiple of 19 based on our 2019 forecast.
We project revenue growth to average about 8% annually between 2019 and 2028. We see the company getting a boost from greater exposure to retail colocation tenants, coming from the 2015 acquisition of Telx. Not only do we think the top line will benefit from the higher revenue per square foot that retail tenants pay compared with the larger-scale customers, but we think retail colocation growth opens the door for many more connections between tenants. Digital Realty now has much more opportunity to connect enterprises to their cloud service providers and other businesses. We forecast interconnection revenue to also average about 8% growth annually through 2028. Rental revenue from leasing space and power will continue to dominate, however, as we project it will still comprise more than 90% of total revenue in 2028. We forecast the firm to add nearly 2 million feet worth of new leases annually for the next five years while maintaining a retention ratio of 80%. We project average rent per square foot to grow 2%-3% per year, leading to rental revenue increasing at a mid-single digit rate annually.
We don't believe the firm has significant opportunities to enhance profitability, especially since we see the increasing power of hyperscale tenants squeezing pricing. However, we think a shift in the mix toward more retail tenants and a higher number of interconnections will result in modest margin expansion. We forecast adjusted EBITDA margin to rise to over 63% by 2028, up 400 basis points from 2018. We expect new square footage growth to slow after the big expansion we forecast over the next five years, so we project capital spending to fall from over 45% of sales in 2018 to just under 20% by 2028.
We rate Digital Realty as a high uncertainty stock primarily because the capital-heavy nature of its business minimizes its ability to be agile in the face of a market downturn. If demand for third-party data centers weakens for any reason, Digital would find itself with too much capacity and a weakened cost profile.
We think technological advances pose the biggest risk to Digital Realty's long-term prospects. The company's future growth depends on cloud service providers and enterprises needing more data storage, compute power, and interconnection as the world becomes more data-centric and connected. However, technological advancement could result in an environment where those increasing needs come to fruition but data centers are not the beneficiaries. Virtualization and Moore's law could result in a scenario where Digital Realty's customers satisfy their exploding data needs with less physical space. Similarly, an increase in the amount of data companies can push through a given fiber connection could result in fewer necessary cross connects for the same amount of traffic. Finally, a greater shift by enterprises toward infrastructure as a service, or IaaS, cloud adoption could reduce their own need for server space, as their data is shifted to cloud providers' servers and stored more efficiently.
Power becoming concentrated in the big cloud providers is the other big risk we see for Digital Realty. Digital's top 20 customers already comprise over half its revenue, and cloud and IT service companies account for 45% of Digital Realty's revenue. We expect their importance to continue growing, leaving Digital Realty more susceptible to these companies playing hardball. We don't think cloud providers want to compete in a retail colocation business, considering the difficulty we believe they'd have in mimicking a similar network of customers, but they could use their leverage to extract concessions.
We assign Digital Realty a Standard stewardship rating. We believe management has done a fine job, but we don't yet see tangible incremental results stemming from management's vision or execution that lead us to rate it exemplary. However, we like the direction the company is going, and we think recent acquisitions (most notably Telx in 2015 and DuPont Fabros in 2017) could potentially be transformative, with more emphasis on retail colocation and interconnection. While the DuPont Fabros acquisition simply netted wholesale data centers and customers, we nonetheless think it will enhance colocation and connections due to its high exposure to cloud service providers. In our view, results over the next few years likely will be largely attributable to management's recent decisions and the risks it took.
CEO Bill Stein has led Digital Realty since 2014 and has been with the firm virtually since inception. He joined the firm as chief investment officer in 2004, the year it went public, and has also served as CFO. Prior to 2004, he was with GI Partners, the private equity fund that previously owned Digital Realty. Since 2015, Andy Power has been the firm's CFO, and we think he is being groomed as an eventual successor to Stein. He previously worked as an investment banker with both Citigroup and Bank of America, where he was managing director of the real estate, gaming, and lodging sectors. He was part of the underwriting team that brought Digital Realty public in 2004 and has been closely involved with the company throughout its history. In 2018, the company changed its organizational chart so that all global sales and marketing functions report to Power. Also part of what we see as a deep bench is CIO Scott Peterson, who has been in the role since 2014 and has been an officer at the company since 2004.
We view Digital Realty's corporate governance as shareholder-friendly. All directors are up for re-election annually, nine of 10 directors independent, and it maintains separate chairman and CEO roles—Laurence Chapman has been chairman of the board since 2017. Named executive compensation seems reasonable, though we think the cash portion of targeted compensation is higher than some peers. Stein's base salary in 2018 was $1 million and represented 9% of his targeted compensation, with 16% being a targeted cash bonus and the remainder in the form of equity awards, which are mostly determined by performance.
We also credit management for its commitment to sustainability. Data centers are among the biggest consumers of power in the world, and Digital Realty has committed to eventually powering its portfolio with 100% renewable energy. In 2016, it procured more than 1.38 gigawatt hours of renewable energy and has 62 certified green buildings globally, both industry-leading metrics. The company has also won numerous sustainability awards and improved its power usage effectiveness, or PUE—an efficiency measurement, by over 20% between 2014 and 2017 in buildings it committed to the U.S. Department of Energy's Better Buildings Challenge.
沒有留言:
張貼留言