Data Centers as Connectivity Providers

It overstates the matter to argue that “interconnection” is the core business of a data center, compared to rack space, power, air conditioning and security at better financial terms than operating one’s own internal data center. 

But it would not be overstating the matter to argue that interconnection is an absolutely vital function of a data center. 

According to Cisco data, most global data traffic actually moves within data centers. In past years, “inside the building data traffic has represented as much as 75 percent of all data traffic.  Another seven to nine percent of global data traffic moved between data centers, according to Cisco. 

source: Cisco 

Nor would it be incorrect to argue that a data center’s core revenues come only partly from real estate value. As it turns out, a growing part of the business is interconnection value.

By some measures, interconnection represents roughly 18.5 percent of Equinix recurring revenues. 

source: Nortia Research, Paolo Gorgo 

source: Nortia Research, Paolo Gorgo 

Other leading data centers also report significant recurring revenues from interconnection services. 

source: Nortia Research

We normally think of connectivity providers as the lead suppliers of “interconnection” or “networking” or “bandwidth,” but data centers also directly earn revenue from supplying interconnection services, access to networks and bandwidth. 

source: Equinix, dgtlinfra  

On its fourth quarter 2022 earnings call, Equinix noted that interconnection business saw revenues for the quarter growing 13 percent,  year-over-year, “outpacing the broader business.”

“Equinix Fabric saw continued growth and is now operating at a $200 million revenue run rate, one of our fastest-growing products,” said Charles Meyers, Equinix CEO.

In Home Broadband, Physical Media Matters, But Perhaps Not as Much as You Might Think

There always is a gap between deployed fiber-to-premises passings and customer uptake. As a practical matter, there is a lag between full marketing results and network construction. Secondly, it does not appear that “fiber access,” in and of itself, necessarily is the preferred consumer choice in competitive markets. 

source: BT, LightReading 

In markets with strong cable operator competition, for example, FTTH tends to get between 40 percent penetration and 45 percent adoption after about three years of marketing. Some FTTH ISPs hope to reach a terminal adoption rate of 50 percent, but that is about the extent of expectations. 

source: IDATE, TelecomTV

Data from other European markets shows similar gaps between facilities deployment and take rates, where take rates hover between 45 percent and 47 percent. And that is a view of physical media choices, not necessarily speed tiers chosen by customers. 

We often assume the key point is service delivery by optical fiber at the premises. The better metric is consumer choice of service plans ranked either by speed or price. 

source: Omdia, Medux 

Customers often choose not to buy the fastest tiers of service on any home broadband network, even if the percentage taking higher-speed tiers continues to grow. That is what one would expect, of course, as “typical” or “average” speeds continue to grow. Still, many consumers choose not to buy the “fastest” tiers, but rather tiers someplace in the middle between fastest and slowest. 

source: OpenVault

The point is that assessing physical media availability does not directly correlate with any particular consumer choices related to speed tiers. Speeds will keep growing on all access platforms over time. So physical media alone does not directly correlate to specific choices consumers make.   

source: RVA, Broadband Communities 

We often equate “fiber access” with “speed.” That requires some qualification. Gigabit or multi-gigabit speeds are possible. That does not mean they are always offered, or that, when offered, that consumers buy the services. 

What matters is both availability and take rates of gigabit and multi-gigabit services. Physical media really is not the issue.

"Sum of the Parts" is Getting More Important

Asset valuation in the digital infrastructure space gets more complicated as firms start to embrace functions across two or more ecosystem roles.

Even within a single category, valuation often requires understanding which "type" of asset is involved.

Infrastructure investors sometimes debate whether “core plus” is a “real” category, but there is no doubt new asset classes that might once have been considered “core plus” now are either “core” assets or, in some cases, “supercore.” The original categories come from the real estate business, and now also are the framework for infrastructure investing generally. 

Those who use the “supercore” appellation essentially move “core” assets to the “supercore” category, while the former “core plus” becomes core. The value-add category then becomes “core plus.” Some add an additional “opportunistic” category including assets in emerging markets and assets whose primary value is asset appreciation rather than operating income. 

source: Mercer 

“Supercore” has in the past referred to assets with regulated prices and rates of return, such as electrical utilities. “Core” assets might be considered those with business moats, pricing power and predictable demand. Toll roads, railroads, energy distribution companies. airports or mobile network towers provide examples. 

Generally speaking, supercore assets are perceived as having the lowest risk, core-plus as having higher risk, greater variability but also higher growth rates or profit margins. 

Key valuation metrics arguably also vary between the asset classes. 

Supercore assets--the most predictable assets, with the highest business moats--often rely on earnings (EBITDA) and pierce-earnings multiples.

Core-plus assets often require injections of new capital, so often rely on internal rate of return metrics. Optical fiber networks often are in this category. 

Core assets might more often be evaluated using net operating income or discounted cash flow analyses. These are considered mature and stable categories. Mobile towers and data centers often are in this category. 

------------------

Virtually all observers would agree that valuation multiples for assets such as data centers are different from valuations of connectivity providers, and that valuations for other assets also will differ from either of those types of assets. 

sources: Oliver Wyman, Bank of America 

Industrial technology, which includes industrial process control, industrial software, vision technology, robots and motion sensing, can feature EV/EBITDA multiples somewhere between “connectivity providers” and data centers. 

source: Capitalmind 

source: Adventis Advisors

To be sure, valuation multiples change over time, and already seems to be true in 2023 for public software companies. 

source: Microcap.co  

Multiple compression from the robust financial markets of 2020 is obvious in every vertical category. 

GICS Sector	12/31/2022	6/30/2022	12/31/2021	6/30/2021	12/31/2020
Communications	8.57	9.27	12.96	14.78	14.14
Consumer Discretionary	14.44	14.98	21.88	23.02	26.09
Consumer Staples	16.42	14.92	17.53	16.53	16.92
Energy	5.37	6.98	8.97	22.52	-
Health Care	16.24	14.93	18.18	19.73	17.36
Industrials	15.06	13.89	17.62	25.12	20.61
Information Technology	15.96	15.91	23.45	22.87	22.65
Materials	9.14	9.05	11.91	15.25	18.18
Real Estate	16.61	17.83	24.48	25.27	21.30
Utilities	13.75	14.35	14.23	13.59	13.63

source: Siblis Research 

The whole point is that industries have different sorts of multiples. And valuation becomes much more complicated when firms start embracing multiple roles within the ecosystem. “Sum of the parts” evaluations then must be conducted to figure out what an asset ought to be worth.

In Business Terms, Platforms Make Money on Transactions, Not Retailing

Platform business models typically involve a few essential roles, according to Angus Ward, Beyond Now CEO. A platform owner orchestrates the “exchange of value” operations. There are “producers” who supply capabilities, “providers”  who own customer relationships,  and “customers” who buy capabilities. 

In many cases, buyers can be sellers and sellers can also be buyers. In some cases the platform might also be a producer of its own branded products. 

source: BeyondNow 

The key point is that the business model is based on facilitating transactions, and acting as a market maker or facilitator, rather than a retailer.

Conglomerate Valuations Always are Tougher than Pure Plays

Fluid ecosystems, where some participants supply multiple types of value, are hard to evaluate, from a financial perspective, especially when some categories of activity have one set of valuation multiples while others have differing multiples. 

That is the reason many observers prefer “pure play” assets rather than conglomerate business models where assets of distinct types are co-mingled. Consider the way some products are sold by multiple entities with different valuation multiples. 

In such cases one often has to develop “sum of the parts” estimates that attempt to account for the different revenue, profit margin and valuation metrics different types of products feature. 

Software-defined interconnection, for example, suggests the changing roles of connectivity and data center providers. In recent years, SDI has been a capability featured by firms such as Equinix for connecting servers that are not colocated within a single data center. 

The idea is to enable network-to-network connections without the need to deploy network appliances at each site, as is required by SD-WAN or MPLS solutions also pitched as “software defined” networking approaches.

But many would argue that software-defined interconnection is about connecting locations, data centers, clouds or apps; colocation facilities; internet exchanges or service providers in a more granular and on-demand way than is possible with SD-WAN. 

Ideally, think about the matter as eliminating the difference between north-south traffic within any single data center and east-west traffic across the globe. 

That essentially means that some managed service providers; some connectivity networks; some data centers and some infrastructure providers also are in the SDI business, even if the specific revenue contributions might be hard to untangle. 

In the meantime, SDI might be considered an emerging business category with contestants in formerly-distinct realms, the most significant perhaps being a function both data centers and connectivity providers supply. 

And leaves analysts with the difficult job of assessing the relative contributions of different product categories--possibly with distinct profit margin and valuation metrics--as drivers of overall valuation.