Five AI trends that are changing the data center industry according to Vertiv

AI continues to drive change in the data center industry, according to a new report from Vertiv (NYSE: VRT), a global leader in critical digital infrastructure. Vertiv Frontiers The report, which draws on expertise from across the organization, describes the technology trends driving current and future innovation, from powering AI to digital twins and adaptive liquid cooling.

“The data center industry is rapidly changing how we design, build and operate data centers to meet the density and speed demands of AI facilities,” says Scott Armul, Product and Technology Manager at Vertiv. We see how technology trends such as extreme densification are driving innovative solutions such as high-voltage DC (direct current) and advanced liquid cooling. These are essential to achieving the gigawatt scale required for AI development. Own energy production and digital twin technology are also expected to contribute to increasing the pace of development.

The Vertiv Frontiers Report builds on the company's previous trend forecasts for data centers. Four important driving forces are identified: increasingly powerful equipment in a smaller space (powered by AI), rapid expansion at gigawatt scale, data centers that function as a cohesive computing unit, and the need to handle increasingly more types of computer chips.

Vertiv identifies five key trends that are now transforming the data center industry:

Power supply for AI

A typical data center today converts AC power to DC power multiple times before it reaches the servers. Each conversion results in energy loss. As AI systems demand more and more power in a smaller space, today's solutions are no longer enough.

New systems with higher DC voltage reduce power losses, require thinner cables and fewer conversions. Power conversion can then be centralized for the entire room instead of each rack. Today, mixed systems are mainly used, but pure DC solutions are becoming more common as the technology matures and power density increases. Local energy production and microgrids are also driving the development towards high-voltage DC.

Distributed AI

The billions invested in AI data centers for large language models (LLM) has created cloud-based services for consumers and businesses. But more and more organizations need to run AI locally instead.

Banks, defense agencies, and hospitals often have to host AI systems in their own facilities. The reasons include legal requirements on where data can be stored, security requirements, or the need for fast response times. This is now driving a wave of upgrades to existing data centers, or new construction, with liquid cooling and power systems that can handle the high energy demands of AI servers.

Increased energy independence

Local, on-site energy production has long been important for independent data centers to ensure operations. Now, increasing problems with access to electricity are driving more data centers to become more energy independent, especially AI data centers.

Investing in your own electricity production, for example with natural gas turbines and other solutions, has several advantages, but is primarily driven by difficulties in obtaining sufficient electricity. Strategies that ”"Bring Your Own Power (and Cooling)"”, where data centers produce their own electricity and manage cooling on site to reduce dependence on the electricity grid, is expected to become part of long-term energy planning.

Design and operation using digital twins

With increasingly powerful AI systems, the need to quickly get these complex facilities up and running increases. By creating digital twins, virtual copies of data centers, the entire facility can be planned before construction begins. IT equipment and infrastructure can be integrated as ready-made modules and put into operation as a unit, halving the time from planning to operational AI systems. This method is crucial for efficiently building the huge facilities that the AI of the future requires.

Adaptive and robust liquid cooling

AI systems have driven the adoption of liquid cooling. But AI can also be used to make the cooling systems themselves smarter and more adaptable. Through enhanced monitoring and AI control, liquid cooling becomes smarter and more robust – the system can predict failures before they occur and optimize the management of fluids and components. This increases reliability and reduces the risk of downtime.

Vertiv Operating in more than 130 countries, the company delivers critical digital infrastructure solutions to data centers, communications networks, and commercial and industrial facilities worldwide. The company's comprehensive portfolio includes power management, cooling, and IT infrastructure solutions and related services – from the cloud to the edge of the network. This integrated approach enables continuous operations, optimal performance, and scalable growth for customers navigating an increasingly complex digital landscape.

For more information about Vertiv solutions or Vertiv's predictions for data center evolution through Vertiv™ Frontiers, visit Vertiv.com.