Latest News

NAIOP SFBA hosted a Data Center Market Overview as it’s March 11^th Zoom-In Webinar. The event consisted of an in-depth discussion on the state of digital real estate and what drives fundamentals in this thriving market sector. Moderated by Michael Hochanadel, Head of Digital Real Estate, Harrison Street, the program featured a candid conversation with four of the top data center deal makers active in the Bay Area market and beyond. Panelists included: Aaron Wangenheim, T5 Data Centers; John Sheputis, Managing Director, GI Partners; Raul Saavedra, Senior Managing Director, JLL Capital Markets; and Mike Armstrong, Managing Director, Stream.

Here is an overview of key points from the discussion:

Q – What makes a market a good market for the data center asset class?

Saavedra

• Generally the business is thought of in regions
• Key attributes are needed to be successful
• Power
• Fiber
• Critical mass of logos
• Big enough MSA
• Incentives
• No accident that certain markets have taken off
• In the West, Santa Clara is primary market
• Located in the heart of Silicon Valley / has power and fiber
• Other areas in the Western Region (that draft off the Bay Area)
• Hillsboro OR
• Arizona
• Utah
• Nevada
• Central Region
• Chicago
• Dallas
• Eastern Region
• Ashburn VA (70% of all internet traffic comes through)
• Has become the haves and have nots with Ashburn and Santa Clara dominating
• A tremendous amount of investment goes in for operators and users

Q - How do you think about investments in the sector? How does having an alternative use case play in?

Armstrong

• Have to ask the question, what could this asset be if and when it’s no longer a data center?
• How can power, connectivity and incentives be redeployed or monetized if your tenant goes away?
• Think creatively about how to repurpose
• In many cases there is not a lot you can do / In some cases there are alternative uses:
• Disaster recovery
• Alternate technology uses
• R & D applications

Q - How do you think about allocating the two different types of opportunities, private equity or venture capital versus the long term mission critical infrastructure perspective?

Sheputis

• Seen tremendous cap rate compression / appreciation of the businesses
• 10 years ago you would be happy to sell at a 9 cap
• When he sold Fortune Data Centers in 2018 they got below a 4 cap
• Now the market understands the business better
• Before, there was little understanding of where the real estate (that is appreciating) stops and where the equipment (that is depreciating) starts
• You don’t see a lot of private equity capital in the large data center transactions
• Too big, too competitive and too much money chasing the sector to make PE returns
• Re-rating of the sector / infrastructure companies have stepped in
• At GI Partners, allocation between PE/Infrastructure and RE is based on what’s being sold
• If it’s a company the infrastructure fund takes a look
• If it’s the deed the real property group takes a look
• The appreciation for the sector is much greater than it used to be
• Everybody got the memo about the sector’s performance especially during Covid
• Many investors are bullish on the sector, but uncertain of how to access anything besides the most core deals, where bidding is very competitive

Q – Is T5 the first mover in what could be trend in the direction of branching off from of investment, ownership, operations and development of data centers into some of the ancillary businesses? It could make it easier for some of the capital to enter the space.

Wangenheim

• There is so much action in the sector now / so many ways you can slice it
• Triple net 20 year shell leases
• Interconnect
• Carrier Hotels
• Large turn key
• There is differentiation and a bit of commoditization (same kit of parts)
• Everyone has to find their niche and figure out where they fit in the ecosystem
• For T5 it was trying not to fall into developer’s dilemma of having to keep developing
• Looked for ways to add value for customers when it doesn’t mean greenfield development
• Customers helped pivot toward this with the direct relationship
• Brought services like operations and construction in house
• Increases competitive edge
• Enhances customer relationships
• Gives more levers to pull on pricing

Q – How is bitcoin impacting demand? What are the other main drivers of demand?

Saavedra

• With bitcoin the credit is unknown / not sure how it gets solved
• Makes it unappealing to operators
• Bitcoin companies typically want a bare bones warehouse with just enough power
• Demand driver is simple: publicly traded cloud service providers (Google, Apple, Amazon, Microsoft)
• Responsible for 80% of absorption
• Reporting 611 MW under construction (twice the number of last year)
• All the operators trying to stay ahead of demand
• Pushing business outside the US (EMEA, Latin America, Asia)
• Doesn’t see it slowing down

Wangenheim

• Remembers the day everyone thought Facebook was risky
• Today ByteDance and other huge transactions in Northern Virginia taking credit risk
• Maybe bitcoin is at the far end of the spectrum
• You don’t always get a Google or a Microsoft so you have to take some bets

Armstrong

• Bitcoin is not manifesting today where there is good overlap with the institutional real estate space
• But don’t sleep on blockchain and the underlying technologies
• As it matures and results in applications with more institutional use it may lead to credit worthy tenants

Sheputis

• Think about who’s doing the absorbing (Google, Microsoft, Amazon, Facebook)
• These companies don’t need anyone’s money
• They’re trying to grow, the speed at which real estate operates is sometimes too slow
• A big deal in Silicon Valley 10 years ago was 6 MW / now they’re an order of magnitude bigger
• No other real estate class has experienced that kind of absorption growth
• Bitcoin is just like any other immature business
• Like gaming companies they can’t make ten year commitments
• But they need a place to go, but given their price sensitivity and uncertain credit, might not be a great lessee for most merchant data center operators

Q – What is the minimum site size appropriate for data center development?

Wangenheim

• It depends on where you’re getting your revenue from
• If it’s purely from the real estate investment you have to go big
• 20 MW to be in the game
• To be competitive and offer the rental rates you have to be at scale
• Same for operations / you can’t operate 7/24 a small data center at the same cost per unit as a big one

Sheputis

• Data center hyperscale business has become highly price driven
• The big developers are being forced to competing on marginal cost, however they can maintain some marge as they innovate on design and build at greater scale, both of which have a deflationary effect on unit rents
• There are other types of data centers that are smaller and proximity based, such as network hubs
• The Edge type for example that are closer to population and there will be a proximity benefit for those

Q – What are your views on yields and cap rates?

Armstrong

• In terms of asset pricing: exit cap rates has seen significant compression over last 7 – 10 years
• Development yields that were in the low teens between 2010 - 2012 now single digits (returns on cost)
• Transaction side seeing similar but less sharp trend / yesterday’s 8 cap is today’s 6 cap (for stable cash flow)
• In data center speak that’s going from an industrial shell to turn key those have come down quite a bit too
• In many cases attributable to the influx of capital coming into the space / not that the market has matured or that people understand the risk and are bifurcating it in a way they otherwise should

Wangenheim

• Trying to get high single digits on greenfield developments
• Biggest assumption is what the exit cap will be
• As everyone knows it could be 8 – 10 years to acquire property and build in phases especially when building at scale

Sheputis

• Even with decreasing yields, developers are getting more clever about how they build and assemble
• The better operators would tell you they’re still making their return targets in part due to better assumptions around exits
• There are a handful of attractive wholesale markets. Silicon Valley is still a very attractive market, but half or two-thirds of the absorption is in Northern Virginia
• Despite the high land costs and more challenging development environment, Silicon Valley still works economically for the cloud providers
• Even at those large scale projects they still benefit from network
• Once someone puts down all the physical connections and the core of their cloud service in one spot it anchors them physically
• They all know that Silicon Valley is expensive but they all still develop and expand here

Saavedra

• Thinks yields will end up being stable at best
• In one campus you are making an investment of $250M – 500M / over $1k per square foot
• The challenge is the restrictions in CA are going to continue to drive costs up
• There are three companies that have so much leverage with operators on pricing
• Unlike other asset classes this sector is very opaque

Q – There are a couple of groups out there pursing ultra large gigawatt projects, many hundreds of acres ideally with direct access to significant power capacity capturing economies of scale for the hyper scale community. What do you think about these opportunities? Are they the wave of the future?

Wangenheim

• The hyper-scalers build their own campuses
• These companies have all the money in the world
• Despite being inefficient and bureaucratic at times they have the expertise
• Unless there is something unique you can bring to the table it’s hard to add value

Sheputis

• It’s fantastic to think about a gigawatt as the unit of absorption
• What’s the developer’s value add? Does Google really want a thinly capitalized developer between them and consequences?
• Even so, the facility is still the small part of the total investment. Think about all the IT equipment that goes into these facilities.
• The developers don’t buy equipment any cheaper vs the big end users. The manufacturers can just as easily sell generators and chillers directly to the small number of companies that do all the absorbing
• The developers value add is bringing speed and performance

Q – What are edge data centers? How to you define the edge? How does it play into the data center market today?

Wangenheim

• It’s what the customer defines it as
• It’s just proximity / availability zones
• Geography seems to be one definer / outside core markets

Q - Do you see any disrupting factors on the horizon that you are worried about?

Sheputis

• The cost of the shell and the network are pretty enduring
• Cap rates for shell will always be better than turn key (shell plus equipment for power, cooling and redundancy)
• Heard all kinds of opinions about how data centers will be designed differently, but the main difference is they are bigger scale, more efficient, and more secure
• Have seen machines get denser in terms of how much heat they reject
• Component wise, the biggest change is in innovation in the design of emergency power systems and the UPS / battery systems

Q - What is area as the potential of being the next Ashburn?

Saavedra

• Likes Hillsboro OR and Utah
• They draft off the Bay Area market and have good underlying real estate
• Hillsboro has hydro power (tech companies love that)

Q - What do you think about antitrust considerations?

Sheputis

• Not worried / it will be noisy but won’t change the rate of absorption
• There are a handful of companies that make all the developers compete
• They also compete with each other
• Everything were doing is enduring to the traffic and growth of IT workloads

Q – What is the impact of Covid? Short term blip or long term trend accelerator?

Armstrong

• Long term trend accelerator
• Thinks in many cases folks moving certain parts of their business to the cloud
• Has caused rationalization and thought about how companies are deploying solutions across their business
• In many cases kick started or accelerated conversations which will manifest in increased technology adoption and usage