Financing data centres in the US and Europe

Overview

Across the world, a large proportion of data centre development funding is reliant on debt.  Debt funds and real estate lenders have long focused on traditional commercial real estate such as offices, living and warehouses. But recently many have been considering financing data centres, in order to spread risk to other sectors and take advantage of a growing asset class.  Data centre tenants are amongst the biggest investment grade companies in the world, with low vacancy and default rates, long-term leases and stable cash flows, so they make for a desirable investment. 

The right lender for any given scenario depends on many factors, including size, the nature and credit worthiness of underlying lease or operator agreements, desired tenure and leverage requirements.  Historically, data centre financing in both Europe and the US has relied heavily on bank loans and syndicated facilities, but as development costs have climbed and refinancing needs intensified, structured finance has started to also appear as a financing tool.

US data centre securitisations

In the US, securitisation as a financing tool for data centres has grown considerably in the past five years.  The US data centre securitisation market is primarily comprised of CMBS and ABS issuances (as further described below).

CMBS

In the US, CMBS of data centres account for approximately 30% of data centre financing deal volume, and this market has been active since about 2020.  A US CMBS involves different layers of rated notes, with the highest class of notes usually obtaining a AAA rating and the rating of the subsequent classes of notes gradually decreasing accordingly.  Data centres tend to be well suited to CMBS due to their high tenant quality and renewal stability.  In the US, the single-asset single-borrower data centre market has been growing in particular recently.

On a CMBS, there tends to be more focus on the underlying real estate (rather than the cashflows or the growth of the business) in comparison with an ABS, but this is combined with a data centre's strong services component, typically a long lease element with high quality tenants (and this contrasts with other types of CMBS which may have less reliable tenants), and strong industry demand and therefore high refinancing likelihood. 

There is less flexibility to add or replace collateral or to upsize on a CMBS.  Hyperscale data centres tend to be funded by CMBS because the cashflows are very stable, there are typically only one or two tenants, and there is little volatility.  A rater of a US data centre CMBS will typically look at the value of the existing bundle of assets and require a high tenant concentration limit.  Helpfully for this structure, a data centre tends to be a long-lived asset.

ABS

ABS represents the other main part of the US data centre securitisation market, and this has been active since around 2018.  Data centre ABS issuance in the US is expected to grow from about $8 billion today to nearly $25 billion by 2028.  US ABS typically have a three-class structure – A, BBB, and BB.  A master trust is put in place, with the equity interest in that trust being pledged in favour of the lender.  The master trust starts off with certain properties and tenants but is continually added to and can double in size during the financing.  Certain conditions precedent must be met in order for additional notes to be issued, such as a rating agency confirmation if the notes are rated.  Eligibility criteria may also apply to new data centres which are added, though these tend to be light.  Additional notes may be issued ranking pari passu, senior or junior to any previously issued notes under the same indenture.  There may be leverage tests to ensure that the new notes do not disrupt the current leverage level (noting that the additional notes may be issued in conjunction with the addition of new assets).

US ABS tend to finance data centres with shorter-term leases and a greater number of tenants, meaning that they are well suited to colocation data centres, wholesale, turnkey, to a lesser extent powered shell, and/or a combination of different types of data centre.  ABS' structural flexibility and scalability makes it well suited for operators pursuing multi-growth strategies and repeat securitisations over time.  The ABS market is particularly favoured by owner-operators, who value the ability to finance growth in phases without restructuring prior deals.  A rater of a US data centre ABS will look predominantly at the cashflow coming from the real estate, although it will also consider the residual value of the underlying real estate as well. 

Some US ABS issuances result from the inclusion in ABS of anticipated repayment dates (“ARD”) as soft maturity triggers (typically at 5 to 7 years).  At the ARD, there is no official maturity date nor event of default for non-payment, but if the debt has not been refinanced, all excess cash flow, after operating costs, interest and reserves, is applied towards principal.  The interest rate also increases, typically by more than 5%.  This means that issuers are strongly incentivised to seek refinancing, to avoid the value of their asset eroding.  Generally, post-ARD interest is paid subordinate in the waterfall (junior to the lowest rated note) and its non-payment is not an event of default – so in reality investors do not expect to receive this additional cashflow and its purpose is more to prevent leakage to the equity.  In contrast to the ARD, the rated final maturity date is normally 30 years, so there is a long tail of around 25 years. 

Features in both types of US data centre securitisation

In both financing structures, there will be regular reporting on cashflows, market data, tenants and the data centre manager, and frequent monitoring of cashflows.  The key difference is that on a CMBS, the sponsor and tenant tend to be household names, whereas on an ABS, the rating agencies will need to do more work to get comfortable, and the operator has a wider scope of work in managing the data centre.  Back-up managers are seen in both cases. 

On both types of deal, triggers for cash traps and manager replacements, such as financial covenant triggers and LTV breaches, are key, though with differing details – for example, on an ABS greater credit can be given to cashflows past maturity, and on an ABS valuations are required more frequently. On an ABS, cashflow mechanics will more likely change during the life of the deal through cashflow diversion triggers.

Other typical features of a US ABS or CMBS are liquidity providers, hard lock boxes (i.e. direct payments from tenants) and central collections accounts to direct cashflows as required, reserve accounts, and a right to replace the operator/manager if warranted.  Most management agreements are structured to expire monthly unless renewed, allowing a further replacement option.  A manager replacement review is typically triggered by a debt service cover ratio (“DSCR”) threshold breach, management default, specified termination event or deterioration in the portfolio's characteristics.  Issuers are required to take out insurance (the benefit of which is given to the indenture trustee during the financing), and environmental assessments and valuations are carried out.

US ABS may have a separate variable funding note (“VFN”) as a feature, which can be drawn during the life of the deal for certain matters, typically capex or construction expenses. The VFN can also finance changes in the data centres during the life of the transaction.

The US securitisation market for data centres is relatively mature in comparison with Europe, particularly in relation to CMBS.   However, because data centres are so new as an asset class, they remain relatively untested.  As such, rating agencies in the US will apply fairly substantial haircuts when modelling and setting their assumptions. 

Non-capital markets issuances

In Europe, in the data centres space, the real estate finance, leveraged finance and project finance markets have until recently been more active than the capital markets.  Particularly significant transactions outside of the capital markets have involved development finance institutions and large data centre companies – for example, the financing by a large multilateral financial institution of €350 million in loans to an Italian telecommunications company to enable it to expand its fibre and data centre footprint in Italy and Greece in 2021, a £700 million syndicated loan by a syndicate of banks to a UK data centre operator in 2022, and a €3.3 billion pan-European financing of data centres owned by a leading European date centre operator in 2025.

The nature of data centres and their role in transferring data lend themselves almost towards being an infrastructure asset than traditional real estate, i.e. more akin to utility projects such as power stations.  Historically, data centres have predominantly tapped the project finance markets. A standard project finance structure features a long-term amortizing bond that fully amortizes over the life of the revenue offtake contract (such as a lease). A newly formed SPV issues a loan via a loan agreement or a bond, out of a discrete trust secured by the cash flows derived from a static single data centre or a small portfolio of data centres.

In both the US and Europe, private credit has been active in the digital infrastructure space, including private equity players investing in or partnering with data centre REITS.  Mid-market lenders have also been raising dedicated funds to finance essential services and infrastructure that enable data centre projects (such as construction and operations). 

In tandem, the loans market for financing data centres continues to be strong.  In this respect, please refer to our prior article, “Beds and sheds – key financing considerations” for details as to features we tend to see in data centre-backed real estate finance loans.  In particular, as noted in that article, protections around the data centre manager tend to be key, both in terms of ensuring it manages the data centre appropriately, and safeguarding against its insolvency or default.  As always, constant availability of the data centre is also given prominence in the documentation, including covenants around connected capacity and disaster recovery services, and customer contracts are subject to some regulation.

The bonds markets

In the past couple of years, we have seen two major public securitisations of the income from UK and German data centres managed by a global provider of data centre campuses. One was a £600 million UK securitisation in 2024 in respect of two data centres located in Wales, UK, and then more recently a €640m German securitisation in 2025 in respect of four data centres located in Frankfurt and Berlin, Germany. Both of these transactions financed hyperscale data centres. The proceeds of the notes were to be applied towards various matters, including corporate purposes and eligible green projects in accordance with its green finance framework.

Due to the energy efficiency of the underlying data centres, the securitisations were both structured and marketed as green bonds and complied with both the ICMA Green Bond Principles and the company’s green finance framework. The German transaction was also modelled to comply with the German Energy Efficiency Act, which contains specific rules for data centres.

The transactions broadly followed the US ABS model. Similar to US ABS and in contrast to CMBS, there was no loan underlying the securitisation, and a long maturity date of June 2050 (traditional non-data centre European CMBS tending to have much shorter maturity dates of around 5 years). They involved insolvency-remote property owning companies at asset level, with cashflows passed up to their parent, the issuer, via intercompany debt, which is also similar to US ABS.

Similar to US ABS, the transactions contained an ARD concept, which distinguishes them further from the European non-data centre CMBS market. On the German transaction the ARD for the Class A1 was relatively early at June 2027, with the ARD for the Class A2 and Class B falling in June 2030. There were similar consequences as in US ABS when the ARD is reached, such as an increased margin. In contrast to US ABS, however, the transactions had shorter tails of 10 and 20 years, rather than 25 years plus, and it was not contemplated that new assets or notes would be added to the transactions. The presence of just a few high-performing underlying assets also made them feel similar to CMBS.

Several other features brought the transactions in line with US ABS, though these could also be portable to European CMBS, such as a separate variable funding note, certain reserves, deferral of interest, debt service cover ratio and loan to value triggers, and cash trap mechanics. As the transactions involved hyperscale tenants, non-disturbance agreements with those tenants were required. Also to ensure continuity, back-up managers were put in place to replace the service manager (which collects rental income and enforces tenant leases) and/or the data centre manager if required. As always, certain protections were included around energy continuity.

Conclusion

With the European data centres market growing significantly, we hope that the two recent securitisations have started to sow the seed for greater capital markets issuances backed by this emerging asset class in Europe and the UK.  This is against a backdrop of existing European data centre financings of different types outside of the capital markets.  Whilst the European data centre securitisation market is still relatively nascent, we would expect it to rapidly grow now, possibly as a diversification play by certain investors alongside the more mature US market.  With strong experience as a firm across the board in relation to this asset class, we look forward to supporting our clients as the data centres financing market continues to grow and innovate. 

Authored by Isabel Tinsley, Sebastian Oebels, Dietmar Helms, Julian Craughan and Emil Arca.