The track record of sponsors and the quality of tenants and their lease terms continue to be the principal credit-risk factors for investors when it comes to debt service, but technological innovation and regulation are emerging risk factors. Also, experienced contractors are vital to managing development and capex factors and minimising construction risks such as cost overruns and delays.

Technological obsolescence

The speed of technological innovation is a risk for data-centre operators and tenants as it threatens to render existing infrastructure obsolete or incompatible with emerging requirements. Many legacy data centres originally designed for simple storage and processing needs already face limitations in scalability or lack of flexibility to integrate new technologies.

The technology refresh cycle for data-centre equipment, during which servers and storage/networking typically require replacement, has a three to seven-year span. That is typically shorter than the anticipated repayment date (ARD) of project debt. Tenants are generally responsible for these updates.

Power and cooling systems, typically the responsibility of data-centre operators, have a much longer lifecycle. Because the ARD of recent data-centre transactions has typically been around five years, debt is likely to be repaid before this critical infrastructure begins to age. For projects with longer lending horizons or repayment at the legal maturity of notes (typically 15-25 years) rather than at the ARD, technological obsolescence becomes a key risk concern for investors.

Advances in high-performance computing, especially for AI training workloads (which teach AI models to recognise patterns and make decisions by processing large datasets) are driving significant changes in data-centre design. These changes involve higher power density, heavier rack loads, and transition from traditional air-to-air cooling to more advanced liquid and other solutions.

Retrofitting older facilities to accommodate these evolving requirements can be challenging and may not always be commercially viable. These dynamics underscore the importance of strategic planning to keep data centres adaptable to evolving standards.

Inference-focused facilities, which handle the process of applying trained AI models to make predictions or decisions, are becoming the predominant use of AI and are typically less power-intensive. New technologies, such as DeepSeek, demonstrate that inference workloads can operate effectively at lower density and on older hardware, helping to mitigate technological obsolescence.

Regulatory constraints

Data centres in Europe face growing regulatory and environmental pressures, including from the Energy Efficiency Directive of 2023, and the EU AI Act of 2024. This has raised operating costs or triggered the need for infrastructure upgrades.

Increasing constraints on energy access is also a significant challenge. Amsterdam, London, and Dublin are already facing power-grid restrictions. Existing facilities may experience stricter energy usage limits, and if they under-use their contracted power, they may risk having their capacity reduced.

Stricter regulations may slow AI adoption or delay deployment, which could lead to lower demand for AI-focused capacity and impact long-term growth expectations for the data-centre sector.

Environmental regulations for data centres are tightening across the EU. The revised Energy Efficiency Directive (EED), effective since September 2023, requires large data centres (above 500kW) to report metrics such as water and power usage, heat reuse and use of renewable energy while adhering to increasingly stringent efficiency standards.

Facilities exceeding 1 MW are encouraged to follow the best practices of the European code of conduct on data-centre energy efficiency, such as efficient cooling, energy monitoring, waste, heat recovery and other measures. Obligations can also arise at the national level. Germany’s Energy Efficiency Act, for example, enforces strict Power Usage Effectiveness targets, mandates minimum heat re-use levels, and promotes full reliance on renewable energy.

Compounding these challenges, the EU AI Act, adopted in 2024, imposes restrictions on high-risk AI applications and requires greater transparency for AI systems and models. While AI act is intended to encourage innovation, it also brings significant compliance costs and administrative burdens.

Factoring in construction risk

The recent data-centre securitisations rated by Scope in the UK and Germany – see below for rating links – have only limited fit-out risk but construction and fit-out risks do need to be factored in. Failure to deliver projects on time generally leads to higher costs and delayed revenue. Tenants can even terminate their leases, although this is unlikely since capacity is typically pre-let to high investment-grade tenants already occupying other data halls in the same facilities.

Source: Scope Ratings

Hyperscale tenants, global cloud providers and large technology companies, are strongly committed to their availability zones deeply entrenched in their infrastructure network. This commitment, together with their significant equipment investment in the properties and the high relocation costs, increase tenants’ propensity not to break the leases even in the event of delays.

See also:

Scope rates German data centre notes issued by Vantage Data Centers Germany Borrower Lux S.à r.l., June 2025

Scope assigns A(SF) rating to the notes of Vantage Data Centers Jersey Borrower SPV Limited, May 2024