
Jason Payne
Data Centre Lead, Real Estate Practice, Marsh UK
-
United Kingdom
Accurately assessing contractual risk between data centre owners and potential tenants is crucial to the long-term viability of a data centre. In this fast-moving industry, even a few hours of outages can result in liabilities of millions of pounds and even trigger contract termination rights.
Involving brokers early in contract negotiations is vital to ensure that contractual liabilities are identified and controlled before terms are agreed. There are many different considerations to bear in mind, based on the type of client the data centre owner is aiming to attract, building design and extent of third-party liabilities, amongst others.
There are likely to be different risks and contractual considerations depending on whether the data centre has a single tenant, such as a hyperscaler (i.e. a very large data and infrastructure organisation with global reach) or multiple, smaller tenants within a co-location data centre. A single hyperscaler client may feel that they have more power to define the terms of the contract. Data centre owners may also feel under pressure to accept those requirements to win business. However, owners need to carefully examine the extent of the liabilities in the contract and whether they are being exposed to excessive risk. If there are multiple suites and tenants within the building, the data centre owner may have more ability to define their own terms, manage liabilities and control contractual risk.
Tenants might want the data centre owner to take on liability for damage to their property whilst installed at the centre. That could expose the owner to additional third-party liabilities that are almost as great as the insurance for the data centre itself, creating a very large exposure that could become difficult to insure. Owners may find themselves buying hundreds of millions of pounds of liability cover to mitigate this risk, when the tenant is usually obligated to insure their equipment themselves. Additionally, Tenants may also be inadvertently opening themselves up to liability for damage to the data centre.
Without this safeguard, the data centre owner could become liable for millions of pounds’ worth of lost revenue from the tenant as a result of an outage, for example.
This level needs to be palatable to the insurance market and the client, as well as being reasonable and enforceable. If a liability cap is poorly worded and is tested in court, it may turn out to be unenforceable. This is particularly relevant in unfamiliar jurisdictions where contracts drafted based on local legislation may not apply to other locations. This is particularly important in contract negotiations for data centres with a single client, where the exposure is higher.
Data Centres are built to be operational almost continuously. Tier four data centres for example are designed to have an expected downtime of only 26 minutes over a year. However, the occupancy contract should be drafted to allow for a minimum of two events over the course of six months. The justification for this is that the service credit mechanism caters for minor outages with termination rights being reserved for systemic critical failures over a longer period of time. This ensures the viability of the data centre business.
Typically, property insurance anticipates that building damage could take weeks or months to rectify, with equivalent loss of rent for that period. But in a data centre, an insured event that causes a relatively short outage might trigger a termination right, resulting in the loss of a year or more’s worth of revenue for the owner. Although data centres are designed for hyper resilience, when something goes wrong, it can be disastrous. Even with a hold harmless agreement and waivers in place, a tenant could have sufficient power to negotiate a settlement or keep the property owner in court, that causes real damage to the data centre owner.
In many traditional PI scenarios, claims typically can take several months if not years to develop and resolve. It is not uncommon for a claim to have a lifespan of 3-5 years from the date of original notification through to closure. For data centres, there are immediate issues that need to be resolved within a much shorter timeframe, often a matter of minutes or hours. PI policies for data centres therefore need to be structured around this need and coverage refined to allow flexibility in claims response and control.
From a PI point of view, data centre owners must make sure that they accurately represent their capabilities when onboarding clients. If owners already have a tenant in mind, they may have specific requirements about the building’s performance levels, or have rigorous environmental standards, for example. But sometimes data centres will be built speculatively and where this occurs owners need to be conscious of not overstating – or understating – the centre’s capabilities when looking for clients. When there is a gap between the service a client is expecting and the output a data centre can provide, this leaves space for Professional Indemnity claims. To understand the risk PI insurers will be interested to learn about the onboarding process along with contractual protections that the data centre owner seeks to impose such as caps on liability.
Making sure data centre owners fully understand contractual risks requires early engagement with an insurance broker who can identify potential problems, share knowledge of other contracts to remove problematic clauses and, design insurance programmes to cover the exposures that cannot be written out.
Marsh brings market-leading knowledge and expertise to managing contractual risk for data centre owners. We’re familiar with where the risks may lie in negotiations and understand best practice across the global data centre market. We can help owners identify what is realistic, achievable and sustainable for the long term.
Data Centre Lead, Real Estate Practice, Marsh UK
United Kingdom
Account Executive, Advisory
United Kingdom