castleforge logo
bg detail

Rising Tides Don’t Lift Boats with Holes: Lessons From Data Centre Distress

A recurring theme in our research letters is the idea that what’s true of something as a whole is not necessarily true of each constituent part. In the context of real estate, structurally challenged sectors typically contain well-positioned subsectors, and even struggling subsectors usually contain individual assets that outperform the wider market.

This idea has a corollary. Just as attractive corners of the market can be found in sectors beset with secular headwinds, sectors supported by tailwinds always contain plenty of losers, including unviable business models, subsectors in structural decline, and simply bad assets.

Over the past year, the data centre sector has supplied a few examples that illustrate how things can go wrong even when you have some long-term tailwinds on your side. Cyxtera Technologies is the latest such case.

Cyxtera is the largest global provider of retail colocation services, operating 65 data centre facilities across 33 markets. The Florida-based company was formed in 2017 via a PE-led carve-out of CenturyLink’s colocation business and, in 2021, effectively went public in a de-SPAC transaction that raised about $650 million. Barely two years later, on 4 June, Cyxtera filed for Chapter 11 bankruptcy.

What happened? Cyxtera’s bankruptcy filing blames surging energy costs and macroeconomic volatility. In truth, however, the company’s problems were rooted in its substantial debt load, which finally became unviable, given a combination of higher interest rates, looming maturities, burdensome capex requirements, and the questionable performance of Cyxtera’s core business.

Cyxtera’s pre-petition balance sheet held about $1 billion of financial debt obligations, all floating rate and all maturing sometime in 2024. As interest rates increased over the last 18 months, Cyxtera’s annualised interest expense on its financial debt reached $76 million in Q1-2023, up from $36 million in Q1-2022.

At the same time, Cyxtera has persistently run free cash flow deficits, largely due to its significant ongoing capex requirements: last year, the company did approximately $300 million of so-called “transaction-adjusted EBITDAR” and spent $214 million on lease payments and $135 million on capex. In short, Cyxtera could no longer service its debt while continuing to pay the rent and fund capital improvements.

The dynamics at play here recall observations we have made in the past about some of the “asset light” flexible office operators, like WeWork. Similar to many of their business models, Cyxtera’s business operates by borrowing long and lending short—specifically, by taking long-term leases on entire data centres and then effectively subleasing capacity at higher rates for shorter terms.

This business model is inherently leveraged. Indeed, even though Cyxtera’s data centres were significantly underutilised, the company generally could not trim costs by exiting underperforming facilities. As a result, the company had little ability to control the timing of its capex: in order to meet rent obligations, Cyxtera needed to lease up vacant capacity, and this required spending money to improve that capacity. Operating leverage and financial leverage can make for a dangerous combination.

Some of Cyxtera’s landlords are now left with a mess. Since Cyxtera’s leases (unlike WeWork’s) were typically signed or guaranteed by the parent, Cyxtera’s landlords were partially shielded from the risks of the company’s leveraged operating model. But only up until a certain point, particularly given Cyxtera’s debt load. The company will presumably now use the bankruptcy process to exit some of its worst-performing and most underinvested-in facilities, leaving their owners with assets that are probably unleasable without substantial upgrades.

This points to a final lesson. A data centre wholly leased to a non-credit operator is not similar to an office or retail building leased to multiple non-credit tenants, where diversification at the building level mitigates some of the tenant-specific risk to the landlord. As a result, the fundamental question for investors in these types of buildings is whether the tenant (or guarantor) will be capable of paying rent for the duration of its lease term, given the various competing claims on its cash over that period, including those stemming from its debt obligations, equipment leases, and capex requirements. Perhaps, in hindsight, an operator like Cyxtera should not have passed this test.

This article first appeared on LinkedIn. To keep updated with our latest announcements follow us on LinkedIn.

9th October 2023

Blog Post