Vertiv outlines AI-driven datacentre infrastructure trends

Artificial intelligence is forcing a fundamental rethink of data centre infrastructure. As AI workloads grow denser and deployment timelines shrink, facilities are being redesigned to operate at unprecedented scale and complexity.

A new report from Vertiv examines how AI-driven data centre infrastructure trends are reshaping power architectures, cooling systems, energy strategies, and operational models across global facilities.

Macro forces shaping AI-era datacentres

According to Vertiv, four macro forces are now driving data centre innovation.

The first is extreme densification. AI and high-performance computing workloads are pushing rack densities far beyond traditional limits, placing new demands on power delivery and thermal management.

The second is gigawatt-scale deployment at speed. Datacentres are no longer expanding gradually. Instead, they are being rolled out rapidly and at a massive scale to meet AI demand.

The third force is the shift towards treating the data centre as a single unit of compute. In the AI era, facilities must be designed, built, and operated as integrated systems rather than loosely connected components.

The fourth is silicon diversification. Datacentre infrastructure must now support a growing range of processors and accelerators, each with unique power and cooling requirements.

Power architectures adapt to AI density

Most datacentres today rely on hybrid AC and DC power distribution, with multiple conversion stages between the grid and IT racks. Vertiv notes that this approach is increasingly strained as AI workloads drive higher power densities.

The report highlights higher-voltage DC architectures as a key response. By reducing current, minimising conductor size, and centralising power conversion at the room level, higher-voltage DC can improve efficiency as rack densities increase.

While hybrid AC and DC systems remain common, Vertiv expects higher-voltage DC adoption to grow as standards mature and on-site power generation becomes more prevalent.

Distributed AI reshapes deployment strategies

Initial investment in AI datacentres has focused on large-scale facilities supporting language models and broad AI adoption. Vertiv suggests that inference delivery models will increasingly vary based on organisational requirements.

Highly regulated sectors such as finance, defence, and healthcare may require private or hybrid AI environments to address data residency, security, and latency concerns. This is likely to sustain demand for on-premise datacentres supported by scalable, high-density power and liquid cooling systems.

Energy autonomy moves beyond resilience

On-site power generation has long supported data centre resiliency. Vertiv points out that power availability constraints are now accelerating a shift towards extended energy autonomy, particularly for AI-focused facilities.

Strategies such as Bring Your Own Power and Cooling are emerging as part of broader plans to ensure sustained capacity. Technologies including natural gas turbines, are being explored to support long-duration on-site generation, driven less by redundancy and more by availability at scale.

Digital twins accelerate design and deployment

As AI infrastructure becomes more complex, deployment speed is becoming critical. Vertiv highlights digital twin technology as a way to virtually design and specify datacentres before physical construction begins.

By integrating IT and critical infrastructure within digital twins, facilities can be deployed as prefabricated, modular units of compute. This approach can significantly reduce deployment timelines and support the rapid gigawatt-scale expansions required for AI growth.

Adaptive liquid cooling becomes mission-critical

Liquid cooling adoption has accelerated as AI hardware outpaces the limits of air-based cooling. Vertiv notes that AI can also be used to further enhance liquid cooling systems themselves.

Through advanced monitoring and control, AI-enabled cooling systems can predict potential failures and optimise fluid management. This adaptive approach is expected to improve reliability, uptime, and protection for high-value hardware and data.

Vertiv’s view of the AI infrastructure shift

Across power, cooling, energy, and operations, Vertiv’s analysis points to a data centre landscape defined by integration rather than incremental change.

Vertiv AI-driven data centre infrastructure trends suggest that future facilities will be planned as cohesive ecosystems, designed to scale quickly while supporting the density and performance demands of AI workloads. For operators, the challenge now lies in managing complexity without sacrificing speed or resilience.