As AI drives explosive growth in computing demand, the data center industry stands at a crossroads between scale and sustainability. Forward-thinking operators are turning to enterprise automation platforms, digital grid management and renewable integration to balance performance, efficiency and profitability in a rapidly electrified world.
By Brett Benson, Emerson
Artificial intelligence (AI) is not just a buzzword, but a technological phenomenon that is a foundational shift in how knowledge, decisions and innovation are created. In industrial operations in particular, the potential for AI to drive improvements has thrust it into the spotlight in a way unseen with any previous technology.
Interestingly, another associated term has begun to emerge into the spotlight in parallel: hyperscale data centers. As AI adoption accelerates, so does the demand for the massive computational power required to drive it. AI is a technology the world expects to be immediately available and accurate — and only a strong infrastructure of data centers can enable that future.
However, while the focus on enormous data centers is somewhat new, the technology itself is not, as technology companies have a long history of building data centers. What has changed is that data centers are now creating greater and more complex power generation supply challenges. Today, the world operates a wide range of different data centers, and the differences among them are typically one of scale and purpose. These factors, in turn, have a significant impact on the facility’s power needs, especially those with the purpose of performing energy-intensive AI training activities as compared to cloud storage.
According to the International Energy Agency, traditional cloud-service data centers — the ones that have been in use for decades — typically operate in the 5- to 10-MW consumption range. In contrast, newer hyperscale data centers designed to run complex and powerful AI training, execution, and other tasks typically consume 100 MW of power or more. Moreover, many hyperscale data centers meet the requirements of the Uptime Institute’s Tier IV standards, guaranteeing 99.995% uptime per year.
Increasing scale creates challenges
As data centers increase in scale and energy use, so too does the value of offsetting part or all their power consumption with renewable energy sources like solar. However, doing so when power must satisfy high availability requirements introduces significant complexity. As smaller sites have evolved over the decades, they have found ways to increase their solar energy production, and many rely heavily on solar energy as part of their energy mix.
Hyperscale data centers, however, are currently under extreme pressure to engineer and build workable, incredibly stable facilities quickly, and as a result, they tend to focus less on renewables as part of their energy portfolio as renewables provide more intermittent power than traditional sources.
Yet, as the earliest projects are completed and hyperscale data centers become more commonplace, that trend is likely to shift, as pressure for incorporation of solar and other renewables to power hyperscale data centers will likely increase as they strive to achieve greater sustainability standards and reduce operating costs.
This means that forward-thinking organizations building and operating either type of data center should be implementing automation technologies designed as part of an enterprise operations platform that provides seamlessly integrated technology from the intelligent field, through the edge, and into the cloud. Such a solution will help future-proof investments, providing more flexibility to shift power generation as new needs emerge (Figure 1).
Figure 1: Implementing a comprehensive, unified automation platform for data center power reduces operational complexity by seamlessly integrating proven technologies and helps future-proof investments by providing more flexibility to shift power generation as new needs emerge.
Incorporating solar can be complex
Many power-hungry facilities want to use solar power because it can be brought online faster than other options and it offers the lowest cost of generation possible when weather conditions are right. However, solar is also the least reliable producer. When the weather is perfect, solar arrays produce energy in abundance to match instantaneous demand or charge battery storage systems. Yet, when the skies are cloudy or dark, solar dramatically underproduces.
The variability of solar, coupled with hyperscalers’ need for high availability, means using large percentages of solar generation cannot be achieved without a powerful control solution built for both visibility and flexibility of operation. Control software will need to connect to both solar and traditional generation assets, while seamlessly integrating with forecasting software to consider current and future conditions based on immediate and predicted weather, loads and asset status.
Modern, power plant automation systems aggregate every traditional and renewable generation asset into a common platform for subsystem interoperability, with a single dashboard providing the required operational visibility.
Instead of jumping between disparate control systems for traditional and renewable assets, operators can instantly see their overall generation capacity, and then make critical decisions to maintain safety, uptime, and profitability. The most advanced systems can also manage the high-speed control necessary for effective generation output control and critical load shedding, helping teams change operations quickly when they do not have enough power from generating assets (Figure 2).
Figure 2: Modern, fit-for-purpose power generation control systems provide operators with a unified view of both traditional and renewable assets—enabling faster, safer decisions that protect uptime and profitability.
The benefits of this comprehensive, unified control system are further extended when a data center adds seamlessly integrated digital grid management (DGM) software. DGM software improves generation planning and scheduling based on weather forecasting and historical load data. At the most basic level, teams can use DGM software to determine future production requirements and make numerous strategic energy decisions. These include day-ahead energy purchases from the utility, rather than purchasing on the spot market, dramatically reducing operating costs. DGM software can also factor in expected derates or asset outages.
DGM software can also help teams better select operational strategies and manage load shedding when a generating asset is offline. The software improves financial decisions, helping teams decide when it is most cost-effective to purchase power off the grid, sell to the grid, or — if the site has hydrogen or battery storage — decide when to use excess solar energy to charge batteries or produce hydrogen.
The key to unlocking these technologies, however, is the ability to integrate them easily and effectively. The best modern technologies are built to seamlessly integrate out of the box, eliminating the data silos that slow operations and limit visibility.
Building for tomorrow
Much of this technology is already in use by forward thinking hyperscalers that are incorporating affordable clean solar energy to help transform smaller data centers into mega-complexes. However, as the hyperscale gold rush subsides, those massive facilities will also turn their sights on solar as a valuable resource to reduce costs, supplement energy needs, and potentially even create new revenue streams.
When they do, the sites that planned ahead and implemented automation technologies built on a scalable, seamlessly integrated enterprise operations platform will find it significantly easier to add the new technologies they need and be among the first to capture competitive advantage.
Brett Benson is the director of global renewable solutions business development for Emerson’s Power and Water Solutions business. He possesses over 30 years of extensive power and automation industry project engineering and management experience with many types of power generators and fuel sources, with a focus on global, strategic renewable projects. In his current role, Brett helps to create value-add, purpose-built strategies for all renewable assets, with a focus on solar PV and battery energy storage. Brett received his Electrical Engineering Degree from Auburn University.
