Demand for hyperscale data centres is surging but what are some of the key considerations for investors when negotiating with a ‘hyperscaler’?

Viewpoints
July 31, 2023
3 minutes

In its Data Centres 2023 Global Outlook, JLL predict that there will be more than 1,000 hyperscale data centre sites worldwide by the end of 2024, compared with around 500 locations five years ago. The US’s dominance of the market is expected to continue but growth is also predicted in the UK, Ireland, Spain and Italy. Whilst the rise of artificial intelligence (AI) will boost demand, it will also provide efficiencies to data centre owners and operators. Further demand is also anticipated as a result of a rise in revenues in the public cloud services market, which is expected to grow by almost 20% between 2022 and 2026.

Whilst growth seems certain, not all of that growth will be from new, ground-up data centre developments. Mergers and acquisitions are likely to feature prominently on the data centre landscape as soaring construction costs prompt investors to find alternative means of growing their market footprint. Cash-rich investors may find opportunity to subsume weaker competitors who are less resilient against economic headwinds and as some data centre providers seek to release locked up liquidity, there is opportunity for investors favouring a sale and leaseback model.

In our experience, hyperscale data centre tenants (being the biggest players in the tech markets, ‘hyperscalers’) will have certain fundamental requirements on structuring and key terms. Many hyperscalers have historically favoured the more traditional structuring model of a master services agreement with ground lease built-in, but we are now seeing a move towards structures which are more akin to a NNN lease arrangement. As the NNN lease is a model more familiar to investors and lenders alike, this shift in approach creates many efficiencies during investment review and financing negotiations.

Given their dominance, hyperscalers continue to drive market terms and have a greater degree of influence over what the investor may do vs. tenants of other more traditional assets. A key example of this is the use of ‘white lists’ and/or ‘black lists’ which allow a hyperscaler to control who an investor landlord may transfer some or all of its investment to at the asset level or in the equity. Equity investors will not be used to restrictions being imposed on its exit and liquidity rights but, when faced with a hyperscaler, this is a hot topic of negotiation. Hyperscalers are, understandably, extremely cautious in ensuring that a competitor cannot inadvertently become its landlord (either directly or indirectly) and the same sensitivity arises in respect of certain debt providers in light of enforcement risk. Such sensitivities may lead to the investor only being granted a limited suite of permitted financing rights that should be carefully considered, negotiated and drafted. These covenants may be particularly tricky to navigate in the current market where traditional lenders are less willing to lend.

Two final hot topics to close – increased sustainability regulation and AI.

Sustainability and energy efficiency are top priorities for data centre users, operators and investors with many self-regulatory initiatives being put in place. Legislation has also been introduced in some regions to control power consumption and development of facilities in certain areas. Further legislation is expected and investors may wish to marry up these requirements with their own ESG covenants set out in investment/operational documents and financing arrangements (following an increase we are seeing in sustainability linked loans). Hyperscalers are increasingly asking for covenants linked to sustainability and energy efficiency, a breach of which may give rise to financial consequences.

AI, whilst bolstering demand for data centres as its use becomes more prevalent, is also increasingly being used for improving performance and efficiency of data centre facilities. Sophisticated operations and power/cooling systems provided by AI can yield savings on power costs.

The proliferation of hyperscale data centres explains why private equity has increased its appetite for the sector (accounting for 91% of data centre M&A in 2022) and why JLL’s Annual Lender Survey predicts data centres to be among the top three asset classes anticipated to see the largest net increase in loan exposure over the next 12-24 months. However, what is also clear to us is that the most successful data centre investors will be those that are alive to the fundamental requirements of hyperscalers and, importantly, how best to navigate them.