Global provider of critical digital infrastructure, Vertiv, is positioning artificial intelligence as a catalyst for a fundamental rethinking of how data centers are designed, built, and operated. In a newly released industry analysis, the company argues that AI-driven workloads, combined with mounting pressure on energy availability and deployment speed, are accelerating a shift toward more integrated, intelligent, and self-sufficient data center architectures.
The report, titled Vertiv Frontiers, reflects a broader industry consensus that the traditional assumptions underpinning data center design are being strained by the rapid rise of AI and high-performance computing. According to Vertiv, the combination of extreme power densities, accelerated deployment timelines, and a more diverse range of silicon architectures is pushing operators to reconsider everything from power distribution and cooling to facility planning and operational models.
Scott Armul, Vertiv’s chief product and technology officer, frames the transition as structural rather than incremental. He notes that AI workloads are forcing the industry to rethink how data centers are conceived, particularly as facilities scale toward gigawatt-level capacity. The speed at which AI infrastructure must now be deployed is reshaping both engineering priorities and operational practices, with technologies such as liquid cooling, higher-voltage DC power, and digital twins becoming increasingly central rather than optional.
One of the most significant pressures highlighted in the report is extreme densification. AI and HPC workloads demand far more power and cooling per rack than traditional enterprise or cloud applications, exposing inefficiencies in legacy hybrid AC/DC power architectures. Today’s designs typically rely on multiple power conversion stages between the grid and IT equipment, introducing energy losses and limiting scalability. Vertiv suggests that higher-voltage DC power distribution, while still emerging, could reduce these inefficiencies by lowering current levels, shrinking conductor sizes, and cutting conversion stages. As rack densities continue to rise, the company expects DC architectures to gain traction, especially as standards mature and equipment ecosystems expand.
The rise of AI is also reshaping where compute happens. While hyperscale data centers built to train large language models have absorbed billions in investment, Vertiv argues that inference workloads will be far more distributed. Enterprises in regulated sectors such as finance, healthcare, and defense often face strict latency, security, and data residency requirements that make fully centralized AI models impractical. As a result, many organizations are expected to deploy private or hybrid AI environments closer to users and data sources. This trend is likely to drive retrofits of existing facilities and the construction of smaller, high-density sites that rely on scalable liquid cooling and robust power systems
Energy availability has emerged as another defining constraint. Historically, on-site generation was primarily a resilience measure, designed to support data centers during grid outages. Vertiv now sees a shift toward greater energy independence as grid access becomes a bottleneck for new capacity, particularly for AI-focused facilities. Investments in on-site generation technologies, such as gas turbines, are increasingly driven by the challenge of securing sufficient grid power rather than by redundancy alone. The report points to a growing interest in “Bring Your Own Power” strategies, often paired with on-site cooling solutions, as operators seek to control their own energy destiny.
Against this backdrop of complexity and urgency, digital twin technology is gaining prominence. Vertiv argues that virtual models of data centers, powered by AI, can dramatically compress planning and deployment timelines. By simulating designs, workloads, and infrastructure interactions before construction begins, operators can reduce costly errors and accelerate time-to-service. In some scenarios, digital twins can cut the time required to deliver usable AI capacity by as much as half. This approach also supports modular, prefabricated designs in which IT and supporting infrastructure are deployed together as standardized compute units, a model the company believes will be essential for gigawatt-scale AI buildouts.
Cooling, long a critical but often overlooked aspect of data center design, is now moving to the forefront. Liquid cooling adoption has surged in response to AI-driven heat loads that exceed the capabilities of traditional air-based systems. Vertiv’s analysis suggests that the next phase of innovation will involve making liquid cooling systems more adaptive and intelligent. By integrating AI-driven monitoring and control, these systems could predict failures, optimize fluid dynamics, and improve resilience, ultimately increasing uptime for costly AI hardware and the workloads it supports.
Taken together, these trends point to a convergence of design and operations. Rather than treating facilities, power, cooling, and IT as separate layers, Vertiv envisions data centers operating as unified systems of compute. This systemic approach reflects the realities of the AI era, where performance, efficiency, and speed are tightly interdependent.
The company’s perspective is shaped by its global footprint and portfolio. Operating in more than 130 countries, Vertiv supplies power management, thermal management, and integrated infrastructure solutions across cloud, edge, and enterprise environments. Its view of the market emphasizes not only technological innovation but also the practical challenges faced by operators navigating regulatory constraints, supply chain pressures, and sustainability goals.
While the Vertiv Frontiers report is inherently forward-looking, it underscores a broader shift underway across the data center industry. AI is no longer just another workload category; it is redefining the physical and operational limits of digital infrastructure. For B2B technology leaders, the implications extend beyond capacity planning to encompass energy strategy, risk management, and long-term architectural choices. As AI adoption accelerates, the ability to integrate power, cooling, and compute into cohesive, adaptable systems may become a defining competitive advantage.
Executive Insights FAQ
What is driving the need for new data center designs?
Rising AI and HPC workloads are increasing power density, heat output, and deployment speed requirements beyond what traditional designs can efficiently support.
Why is higher-voltage DC power gaining attention?
Higher-voltage DC can reduce energy losses, simplify power distribution, and better support extreme rack densities associated with AI workloads.
How does AI influence data center location strategies?
Inference workloads often need to run closer to users or sensitive data, pushing organizations toward distributed, on-premise, or hybrid AI deployments.
What role do digital twins play in AI infrastructure?
Digital twins enable virtual planning and optimization of data centers, reducing deployment time and supporting modular, scalable designs.
Why is liquid cooling becoming essential?
AI hardware generates heat levels that exceed air cooling limits, making liquid cooling necessary for performance, reliability, and future scalability.
