Digital TransformationEnviromental

Sustainable Power: Solving the energy challenges of AI

By Mark Yeeles, VP, Secure Power division, Schneider Electric UK&I

Today, many of the macro trends accelerating data centre adoption are being driven by increasingly compute intensive applications. Generative artificial intelligence (GenAI) and AI workloads, large scale data analytics, and accelerated computing are all impacting demand for physical space, connectivity, power and cooling.

In certain geographies such as Dublin and London this has already led to constraints, and now, both government and industry are working to solve the challenges.

Energy demand

There is no simple answer to the complex issues surrounding power. However, a careful strategy built around the most advanced and energy-efficient equipment, AI-enhanced remote monitoring and predictive analytics, and underpinned by innovative energy systems will not only ensure continued, uninterruptible power, but provide a sustainable foundation for the data centres and grids of the future.

According to the IEA, electricity consumption from data centres, AI and cryptocurrency processing could double by 2026, and data centres are currently said to account for around 1% of global electricity consumption. In larger economies such as the US, China, and the EU, data centres account for around 2-4% of total electricity consumption, and in Ireland it could be over 20%.

This level of growth has had a measurable impact on emissions too, with some hyperscalers experiencing a carbon footprint increase of 30% due to indirect emissions from data centres.

Geographical impact

Once seen as a centre for data centre excellence, growth and development, Dublin is in a challenging position. A combination of pressure on the national energy grid and ambitious renewable energy targets has effectively halted new data centre connections since 2021.

Whilst the consultation process led by the Commission for the Regulation of Utilities (CRU) is ongoing, what’s clear is that greater collaboration between government and industry is needed to solve both the data centre, energy and sustainability challenges facing the country.

Similarly in London, a recent report has found that there are up to 400GW of grid requests that may be effectively holding up data centre developments, the strong majority of which (70%) may never see approval. The UK National Grid CEO John Pettigrew has said that “bold action” is needed to create a network able to cope with growing demand from AI.

A new approach to energy generation

Despite an apparent similarity of situations, the two geographies are taking very different approaches. In Dublin the constraints on energy availability has led the CRU to propose strict new rules for energy generation, where new facilities connecting to the grid will be required to provide generation and/or storage capacity to match their requested demand.

Additionally, these new distributed energy systems would be required to participate in the electricity market, improving the resiliency and reliability of the grid, while reporting on their annual consumption of renewable energy and their associated CO2 emissions.

In the UK, encouraged by the measures set out in the UK Government’s AI Opportunities Action Plan, operators are beginning to buildout in a host of new locations, including the North of England. In Greater Manchester, for example, Kao Data has been working with local government to bring its £350 million facility to life, while Microsoft’s £100 million facility in Leeds brings with it new significant advantages for the Northwest.

In Wales, Vantage’s multi-billion pound Bridgeport development may also become one of Europe’s largest data centre campuses and the company plans to invest over £12 billion in data centres across the UK – creating over 11,500 jobs in the process.

Many of these facilities are likely to deploy renewable energy sources (RES), while using innovative energy infrastructure such as battery energy storage systems (BESS), high-density Uninterruptible Power Supplies (UPS) and distributed energy resources to accelerate both their infrastructure deployments, while overcoming energy constraints.

Energy independence

Due to widespread issues of energy constraints, data centre operators are increasingly planning for greater energy independence. Instead of having the ability to operate for just a few hours the event of a power outage, they’re now looking to technology to enable extended periods of self-sufficiency.

This is supported, in part, by the development of new UPS equipment to provide battery backup and advanced power protection. The latest technologies, for example, feature ultra-compact, pioneering high-density designs, and fault-tolerant architecture that can maximise availability, while delivering up to 99% energy efficiency.

The ultimate expression of this growing need for energy independence can be found in microgrids – small-scale power systems that operate independently of the grid and generate electricity for a host of use-cases including university campuses, hospitals, industrial manufacturing systems and data centres.

Additionally, microgrids can work in conjunction with traditional power grids, or function autonomously, delivering power from distributed energy resources (DERs) such as solar panels, wind turbines, fuel cells and energy storage systems to exactly where it’s needed. Furthermore, they can provide additional grid support by providing energy balancing capabilities due to the variability and intermittency of large scale RES.

Small Modular Reactors (SMRs)

With energy challenges continuing apace, some operators are exploring nuclear power generation. Amazon, Google, and Microsoft, for example, have all engaged in direct partnerships to either situate infrastructure near to these systems, or have made commitments to consume energy from existing nuclear power sources.

Another direct avenue for the nuclear path, however, is the development of small modular reactors (SMRs). Based on decades of experience with applications as submarines and icebreakers, there are now around 80 development projects around the world looking, generating capacities of 300MW per system up-to around 500MW.

This new generation of SMRs are based around technologies that are safer and produce less waste, while enjoying a high degree of recyclability in the fuels used. Designs employing technologies such as molten salt, Thorium cycle, and other non-traditional approaches mean safer, more reliable reactors, and are seen as ideal for high demand data centres, and as a basis for microgrids – reducing energy losses, and virtually eliminating the emissions associated with power generation.

Further, all of these innovative energy solutions are being combined with AI-enhanced management systems and predictive analytics built on decades of experience in managing complex networks. As such, they allow data centre operators to meet the challenging demand of the AI boom, with high-density and sustainable power solutions, while supporting national grids to develop and bring on more renewable energy in time.

With innovative technologies now providing a vehicle for greater energy independence, the data centres of the future can provide a platform for the responsible and sustainable growth of AI, while playing a critical role in advancing the UK and Ireland’s digital economies. Further, by harnessing the power of the ecosystem, data centres can be a catalyst for the green transition, and act as a fourth utility that accelerates the development of renewable energy.

Author

Related Articles

Back to top button