As AI puts growing pressure on data centre power supplies, Brandon Spencer, President, Motion Business Area at ABB, argues that making better use of existing capacity could be just as important as generating more of it.
The race to scale artificial intelligence has sparked a global debate about power. Can grids keep up? Will energy supply become the bottleneck? How quickly can new generation capacity come online?
These are important questions. But they risk overlooking a more immediate opportunity.
The next phase of AI growth will not be determined solely by how much electricity we can generate. It will depend on how intelligently we use the electricity we already have. For data centre operators, the challenge is becoming clear. As AI workloads increase, the IEA predicts that global data centre electricity use is on track to nearly double by 2030. Facilities must deliver unprecedented levels of computing power while maintaining reliability, controlling operating costs and meeting increasingly ambitious sustainability targets. In this environment, success is no longer just about access to power. It is about infrastructure.
Reliability, efficiency and scalability must work together
For years, data centre infrastructure was designed around predictable growth patterns. AI has changed those assumptions. Higher rack densities, more demanding cooling requirements and growing expectations around uptime are pushing operators to rethink how facilities are designed and operated. The stakes are significant: downtime costs average $5,600 per minute, and over 60% of failures result in at least $100,000 in losses.
Traditionally, reliability, efficiency and scalability were often treated as separate objectives. Today, they have become inseparable. A reliable facility that is inefficient will struggle with rising energy costs. An efficient facility that cannot scale will quickly become constrained. And a scalable facility without resilience risks costly interruptions. The challenge is no longer choosing between these priorities. It is achieving all three at the same time.
Every megawatt matters
Much of the industry’s attention has focused on securing additional electricity supply. Yet one of the fastest ways to increase available capacity is to use existing power more efficiently. Smarter cooling systems, high-efficiency motors, variable speed drives, intelligent controls and predictive maintenance technologies can all reduce the amount of electricity required to perform the same computing workload. In fact, some estimates show that predictive asset management can decrease maintenance time by 30% and reduce operating expenses by 40%, with potential overall maintenance cost reductions reaching 85%.
While individual improvements may appear incremental, their combined impact can be substantial. Energy efficiency reduces operating expenditure, improves resilience and frees up valuable electrical capacity. In practice, every megawatt saved is a megawatt that can be redirected to support growth. For operators facing power constraints, efficiency should not be viewed simply as a sustainability initiative. It can also support future growth.
Infrastructure decisions are becoming business decisions
As AI adoption accelerates, infrastructure decisions are becoming business decisions. The operators best positioned for long-term success will be those that can build facilities capable of scaling quickly while maintaining performance and uptime. That requires a more integrated approach to power architecture, cooling, automation and digitalisation – technologies that must work together seamlessly.
Standardised modular infrastructure, for example, can help reduce deployment times, allowing operators to bring new capacity online faster while reducing engineering risk. When operators look beyond individual components and focus on how systems work together, from grid to rack, they can not only consume less energy but also extract more value from every kilowatt-hour and accelerate time to revenue. This is changing the role of infrastructure from a supporting function into an increasingly important part of wider business strategy.
Building the foundation for the AI economy
Data centres have become critical infrastructure for modern economies, supporting everything from digital commerce and financial services to healthcare, manufacturing and artificial intelligence. Their continued growth is essential.
Many of the technologies required to support that growth already exist. Integrated approaches encompassing advanced electrification and cooling systems, high-efficiency motors, drives and digital technologies can all support the AI economy’s next phase of growth. These technologies give operators more options to build and scale facilities that are both reliable and efficient. The challenge is therefore not only developing new sources of power, but deploying existing technologies pragmatically and making better use of the capacity already available.
The path forward
The future of AI will not be determined by compute power alone. It will also be shaped by the quality of the infrastructure behind it. For data centre operators, the takeaway is straightforward: meeting future demand will not simply be a question of consuming more energy. It will also depend on using energy more intelligently.
As pressure on grids and available capacity increases, smarter infrastructure will have an increasingly important role to play in determining how far existing power resources can support the next phase of data centre growth.
