With grid connections becoming a major constraint on new data centre development, on-site generation is increasingly being considered for more than emergency backup.

In this sponsored In the Spotlight interview, Data Centre Review speaks to Peter Glenn, Commercial Director at RSC, about the options available to operators, the trade-offs between gas engines, turbines and fuel cells, and how developers can meet immediate power requirements without closing off longer-term sustainability options.

Prefer to watch the video? You can catch the full interview here. 

Why has power become such a critical issue for data centre developers, particularly as AI infrastructure continues to grow?

Peter: The rapid growth of AI is increasing electricity demand faster than grids can add capacity. In many cases, securing land and servers is easier than securing a timely grid connection.

Developers can face connection waits of many years, while the commercial window for bringing new computing capacity online is generally much shorter. That is changing how power infrastructure is considered.

On-site generation is no longer viewed only as emergency backup. It can provide bridge power while a grid connection is completed, or supplement a constrained connection.

The important point is that data centres do not simply need more megawatts. They need dependable power that can be deployed quickly, respond to changing loads and continue operating when individual generating units are being serviced.

That naturally brings modular technologies into the conversation. A plant assembled from multiple gas engine units, for example, can be expanded piecemeal rather than requiring the entire final capacity to be installed on day one.

There are several technologies available for on-site generation. What should operators consider when comparing gas engines, gas turbines and fuel cells?

Peter: There is no single superior technology. The right choice depends on the site, operating profile and delivery timescale.

Gas turbines can achieve excellent efficiency and power density, which makes them attractive for very large, steady baseload requirements. However, they are relatively complex and can face long delivery timescales.

Fuel cells can provide quiet operation and very low local air pollution, but they generally carry higher capital costs and cell replacement times can be difficult to predict.

Gas engines tend to occupy a useful middle ground. They are readily available, offer very good efficiency and rapid response, and are modular.

That modularity is important because capacity can be commissioned in stages. One engine can also be maintained without taking the whole plant offline, which can be a significant benefit.

AI workloads can be much more dynamic than traditional data centre loads. How does that changing demand profile affect the choice of generating technology?

Peter: AI data centres can create highly variable electrical loads, with power demand rising and falling quickly to match the workload.

Generation designed mainly for steady baseload operation is not necessarily optimised for that environment. Operators therefore need to look at start-up times, ramp rates and the number of starts a plant can accommodate, rather than simply maximum output.

Gas engines can reach full load quickly and, because they are modular, control systems can automatically add or remove units to match the load.

That flexibility, particularly when coupled with batteries, can offer greater resilience and generally better efficiency.

Data centres are also being developed across a much wider range of climates. How much difference can temperature, humidity and altitude make when selecting a power solution?

Peter: Potentially a great deal.

The rated output shown on a specification sheet is normally based on standard conditions, but an actual data centre may be operating in desert heat, tropical humidity or at altitude.

Gas turbine output can fall as air temperature or elevation increases. Some projects compensate with additional equipment or water injection, but that introduces more infrastructure and, importantly, more cost and water demand in water-stressed locations.

Gas engines do not suffer from the same severe loss of efficiency at high temperatures or altitudes as gas turbines, and they are generally much more resilient to changes in ambient conditions. That can potentially make them a better solution in varied climates.

Operators are under pressure to solve immediate power constraints while still meeting longer-term sustainability commitments. How can they make a near-term decision without closing off future options?

Peter: In my view, the best approach is to avoid treating today’s solution as a technology dead end.

Operators and data centre owners should examine efficiency and emissions controls, but also whether the plant can adapt to different fuels and work alongside renewables, storage and a future grid connection.

Modern gas engines can potentially support combinations of natural gas and biomethane and, depending on the specific equipment, hydrogen blends or even renewable synthetic methane.

Emerging options such as geothermal and fusion may eventually play important roles, but most data centre projects cannot base a current delivery commitment on technologies that are not yet commercially available at the required scale.

So the practical question is what can deliver reliable capacity within the required timescale while preserving future choices.

In most cases, a gas engine solution deserves serious consideration because it brings together speed, flexibility, redundancy and a credible pathway for fuel evolution.

Gas engine waste heat can also be captured and converted into cooling through the use of absorption chillers. Coupled with chiller heat pumps, this can meet entire cooling loads, removing the need for water and improving engine efficiency.