2018 may go down in history as the year of the great cloud computing explosion. Spending on cloud infrastructure has grown rapidly over the past five years, but in 2018 it leapt higher amidst a spending spree that re-shaped the entire cloud landscape. Facebook’s (FB) capital expenditures (capex) doubled from $6.7 billion to $13.9 billion. Google (GOOG) increased their capex spend from $13 billion to $25 billion. Microsoft’s (MSFT) capex leapt from $8.6 billion to $14.2 billion. If you include Apple (AAPL) and Amazon (AMZN), aggregate capex spending for these 5 tech giants increased from $52 billion to $80.6 billion, a $28 billion increase, which represents year over year growth of 54%.
That’s $28 billion with a B. This spike in capex spending represents the view these companies have of the future of digital data processing and storage, and it required a huge amount of new data center infrastructure to support. It’s no surprise then, that data center absorption increased to 292 megawatts (MW) in the first half of 2018 according to JLL, a growth in leased MW’s of 31.9%. The majority of this new data center space was booked in northern Virginia, the unofficial internet capital of the world. Leasing in this market was 168 MW in early 2018, or 57% of total leasing over that period.
With such a banner year in data center leasing on the books, one would imagine that the public data center REITs would have one of their best performance years of all time. This, however, was not the case in 2018. The data center REITs returned -14.5% in 2018, their worst performance year since 2013. Uhhh…what? This conundrum has left many data center investors scratching their heads, as record leasing has led to poor returns.
So what’s to blame? Did the public companies somehow not keep pace with the rest of the market from a leasing perspective? Using data through Q3 of last year, the public data center REITs (DLR, CONE, QTS, COR) had leased 2.14 million square feet of new data center space over the previous four quarters. This compares to 1.3 million square feet for the same statistic from one year earlier, a growth rate of 55%. This would indicate that data center leasing grew at about the same rate as cloud spending for the public data center REITs. If leasing was not the problem in 2018…what was?
The lack of performance in data center REITs last year has much more to do with expectations that actual results. Expectations are always the wild-card of the stock market. Bad results can be good for a stock if expectations are low enough, and similarly good results can be bad for a stock if expectations are too high. In early 2018, it’s clear in retrospect that analyst expectations for 2019 funds from operations (FFO – a measure of REIT earnings) were much too high. Over the course of 2018, FFO estimates for QTS, CONE, COR, and DLR decreased by about 10% on average. This was despite a record leasing year for the space as detailed above.
A number of company specific issues were responsible for this expectations re-set. Digital Realty (DLR) bought a large FFO dilutive Brazilian data center platform called Ascenty, CyrusOne (CONE) bought a large FFO dilutive European data center platform called Zenium, QTS Realty Trust (QTS) sold its C3 business, and Coresite (COR) saw delays in delivering new leasable data center space. Couple these events with tentative comments from Amazon on its Q3 earnings call, and the result was 10% reductions in 2019 FFO estimates.
The question then becomes whether more negative headlines lie on the horizon. To find some clues we can start with the large cloud player who’s 3Q earnings call comments fueled analyst negativity surrounding future cloud spending, Amazon (AMZN).
Amazon is a heavy capex spender in the cloud space, with 2018 capital expenditures of $13.4 billion. Amazon’s capex was up 12% in 2018, which was a disappointment to analysts. During the company's Q&A on October 25th of 2018 CFO Brian Olsavsky said the following:
“If you look at capital leases, which is where we spend money for the data centers, it's up only 9% year-over-year, trailing 12 months, and it was up 69% last year, at the end of the year.”
This comment along with less than inspiring earnings from DLR sent the data center REITs down about 7% from October 25th to the 27th of last year. The narrative that emerged was that “cloud spending is slowing”. This in theory would lead to less leasing and slower growth for the data center REITs.
But did analysts give these comments too much weight? Let’s see what Amazon had to say on their fourth quarter call which occurred on Jan 31st of 2019.
“I see total CapEx, that was, grew 33% in Q1, 1% in Q2, negative 1% in Q3 and then positive 17% in Q4. Just the quarter itself. So there was as you say a bit of investment in Q4 relative to Q2 and Q3. I still think the 17% is a low number for us when you talk about supporting the AWS business that's still growing in a very high clip… I would say that I would consider 2018 to be a lighter investment year and a lighter year for adding fixed head count.”
This commentary paints a much rosier picture than the previous narrative surrounding cloud spending. After ramping spending on cloud in 2017, Amazon seemingly took time to digest these purchases in 2018, with two quarters of flat growth in 2Q and 3Q. By 4Q, however, they had ramped spending back up, growing 17%, and indicating it will continue to be elevated in 2019.
The broader point here is that capex spending is un-predictable and should not be extrapolated linearly in either direction. Will cloud spending continue to grow at 50% per year into the visible future? No. Facebook and Google are not likely to grow capex at 107% and 91% in 2019, but that does not mean that leasing will suddenly dry up for the data center providers. Amazon is set to re-accelerate capex, while others may digest current inventory, and data center leasing is likely to continue at a strong clip.
Commentary from the brokerage community echoes this sentiment. A recent article from DataCenterFrontier.com quoted Allen Tucker of JLL as saying “I don’t think it’s slowing down…we’re seeing tremendous pent-up growth and demand.” Later in the article Sami Badri of Credit Suisse is quoted as saying “Nothing has changed in the last three months for hyperscalers.”
Looking at data center multiples, you would not guess that this was the case. Relative to consensus 2020 EBITDA estimates, the data center sector looks extremely inexpensive relative to other REITs. On a market cap weighted basis, data centers trade at a 15.6x multiple of expected 2020 EBITDA. This compares to apartment REITs at 20.7x, warehouse (industrial) REITs at 22.8x, and the cell-tower (infrastructure) companies at 19.5x. The data centers in fact trade in-line with the shopping center REITs using this metric, which is ironic considering Amazon is leasing space in data centers, while simultaneously pulling demand out of shopping centers.
* Source: Bloomberg LP and Serenity Alternative Investments
The sectors valuation has also fallen relative to history. Going into last year, the data center REITs traded at an average FFO multiple of about 20.4x. As of this writing that multiple has compressed to 17.5x, illustrating the drop in expectations that occurred during 2018.
In addition to being extremely cheap, the publicly traded data center REITs are likely to capture a higher share of leasing in 2019 than they did in 2018. Part of the reason FFO estimates had to come down in 2018 was that some of the REITs simply ran out of space to lease. DLR spent most of 2018 at over 93% leased in northern Virginia, the highest absorption data center market. Coresite similarly had to buy a large land parcel to continue developing in northern Virginia, which takes time to build out and is completing over the course of the next 3 quarters. With more space available in the busiest data center market in the country, the REITs are much more likely to hit estimates this year and possibly even exceed them.
In the end the data center REITs are still tied to one of the fastest growing pockets of the U.S. economy. Every day more companies move applications into the cloud, deploy new computationally intense machine learning applications, and prepare for the data deluge that will be generated by new IOT devices. Development yields for data centers still exceed those of other major property types, and EBITDA growth could be the highest in the REIT universe in both 2019 and 2020.
With the sector trading at a wide discount to consensus NAV, and a large discount to peers based on 2020 EBITDA estimates, we think it is poised for a rebound after a year of sub-par returns. Assuming 10% NAV growth (in-line with history) and 10% multiple expansion (closing part of the gap to NAV and peer multiples), the possibility exists for 20% returns over the following 12 months.
Risks to the space include private data center providers over-building, continued dilutive acquisitions, or an economically induced slow-down in cloud spending. Put simply, supply would have to increase meaningfully, or demand would have to slow meaningfully in order for data centers to see slower growth than what is currently priced into the market. While these are both certainly possibilities, brokers have seen little evidence of a slow-down in demand, and most of the supply that has been introduced to the market has been pre-leased to this point. Data center management teams have set conservative expectations for leasing so far in 2019, but there is always the possibility of them falling short.
On balance we believe too much negativity is priced into a sector that has a history of exceeding expectations. As data demand continues to grow and the industry continues to mature, investors will have to decide if data centers still deserve to trade at a multiple discount to peers. We think the combination of higher growth and discounted valuations will be too sweet to pass up.
Jump on the data train, not in front of it.
Disclosure: I am/we are long DLR, COR, QTS, SWCH.
Additional disclosure: All pricing information sourced from Bloomberg, LP and publicly available company documents and SEC filings.
The information contained herein does not constitute an offer to sell or the solicitation of an offer to purchase any security or investment product. Any such offer or solicitation may only be made by means of delivery of an approved confidential offering memorandum and only in those jurisdictions where permitted by law. Prospective investors should inform themselves and take appropriate advice as to any applicable legal requirements and any applicable taxation and exchange control regulations in the countries and/or states of their citizenship, residence or domicile which might be relevant to the subscription, purchase, holding, exchange, redemption or disposal of any investments. The information contained herein does not take into account the particular investment objectives or financial circumstances of any specific person who may receive it. Before making an investment, prospective investors are advised to thoroughly and carefully review the offering memorandum with their financial, legal and tax advisers to determine whether an investment such as this is suitable for them.