Vertiv Outlines Data Center Evolution and AI Infrastructure Strategy

Exclusive Interview with Peter Lambrecht, Vice President Sales, EMEA, Vertiv

What specific challenges do clients face in powering and cooling AI infrastructure, and how did you address these at GITEX this year?

At GITEX this year, the focus on artificial intelligence (AI) was unmistakable. Every booth showcased AI-driven solutions running on GPU-powered servers from leading companies like NVIDIA. However, for AI applications to function effectively, the right infrastructure is essential to power and cool these GPUs, ensuring smooth and efficient performance. For us, this highlights our expertise in AI infrastructure, designed to support these platforms optimally.

One of our key focus areas is liquid cooling, as traditional air cooling in data centers is no longer sufficient. With rack densities now reaching 50, 60, and even up to 132 kilowatts per rack, air cooling alone cannot handle the thermal load. Liquid cooling has become critical, efficiently drawing heat away and directing it elsewhere. This core technology, developed with our partner NVIDIA, supports the deployment of GPUs worldwide, and together we provide, market, and deliver these advanced solutions to our clients.

Cooling, however, is only one part of the equation. The shift to AI also requires a comprehensive approach to power management, as AI workloads significantly alter electricity load patterns. Our solutions are designed to meet these power demands, and we have showcased these capabilities at our booth. We’ve been actively engaging with partners and clients to address these challenges as they implement their AI solutions, ensuring that both cooling and power needs are effectively met.

Can you elaborate on your partnership with NVIDIA and how it has evolved over the years?

Our partnership with NVIDIA has grown significantly over the years, and over the past year, it has reached an unprecedented level of collaboration. Both of our CEOs, Jensen Huang of NVIDIA and Giordano Albertazzi of Vertiv communicate regularly, aligning closely on joint development initiatives. Together, we create environments optimized for NVIDIA’s cutting-edge chips, ensuring they have the necessary infrastructure to operate at peak performance.

In a recent joint announcement, both CEOs unveiled new solutions that integrate NVIDIA GPUs with Vertiv’s advanced infrastructure, designed to maximize efficiency and reliability. This collaboration represents the core strength of our partnership, where we have refined reference designs that allow NVIDIA to deploy their GPUs seamlessly and effectively on a global scale.

What current trends do you see in the data center and critical infrastructure market? With many hyperscalers entering the market, what is your perspective on this development?

The data center market is dominated by Hyperscalers, whether through their direct deployments or co-location facilities—two primary models for quickly scaling data center capacity. These Hyperscalers are making substantial investments in infrastructure, fueling competition in this fast-evolving landscape.

The AI race is fully underway, with industry giants all striving toward the same goal. As their infrastructure partner, we are advancing with them, providing the essential support they need to drive this innovation. At the same time, the growth of the cloud sector remains foundational and continues to expand robustly. What we are seeing now is a dual growth trajectory: traditional cloud business growth compounded by the accelerated demand for AI infrastructure.

Trends in data centers reveal a marked increase in power consumption. A few years ago, a five-megawatt data center was considered significant; soon, however, a five-megawatt capacity will fit within a 10×10-meter room as rack density skyrockets, reducing the need for extensive white space but requiring expanded infrastructure areas. Data centers are scaling up to unprecedented sizes, with discussions now involving capacities of 300-400 megawatts, or even gigawatts. Visualizing a gigawatt-sized facility is challenging, yet that is the direction the industry is moving—toward ultra-dense, compacted facilities where every element is intensified, driving an enormous need for power.

As data centers continue to grow, how do you view the sustainability aspect associated with these large facilities?

Today, nearly everyone has a mobile phone in hand, yet data centers—the backbone of our digital lives—often face criticism despite being indispensable to modern society. Data centers are not disappearing; on the contrary, they are set to expand as the pace of digitalization accelerates globally. Power generation remains, and will continue to be, a critical challenge, particularly in regions where resources are limited.

Currently, we see significant advancements in AI infrastructure across Northern Europe. Countries like Sweden, Finland, and Norway benefit from ample hydropower and renewable energy sources, making them well-suited for sustainable AI development. Meanwhile, the Middle East is experiencing a technology boom, backed by its rich energy resources and favorable conditions for large-scale investment in data centers.

There’s also a rising trend toward on-site power generation, with organizations increasingly considering dedicated power stations and micro-grids that tap into renewable or alternative energy sources. In the U.S., for instance, discussions are underway about small, mobile nuclear reactors to support local power needs. Finding sustainable power solutions has become imperative. A sudden surge in electric vehicle usage, for instance, could stress current power supplies dramatically, underscoring the need for substantial changes in our energy landscape.

What support and services does Vertiv offer to its customers? What types of infrastructure investments is Vertiv making in various parts of the world?

At Vertiv, service is arguably the most critical aspect of what we do. Selling a product is only one phase of its lifecycle; the real value lies in our ability to maintain and support that product for the next 10 to 15 years. The availability of highly skilled labor and expertise is essential, as we know that issues will inevitably arise. This is why resilience is a cornerstone of data centers. A strong service infrastructure is vital for addressing challenges promptly when they occur. Just as a car will eventually break down, data center systems too will face difficulties over time. At Vertiv, we have developed an exceptional service framework to ensure that we are always prepared to support our clients.

My philosophy is simple: we don’t sell a product unless we can guarantee the service to support it. When you sell a solution, you’re essentially selling a potential future problem, so ensuring your service capabilities are in place is vital for sustainable growth. This commitment to service is one of our key differentiators in the market.

We are continuously enhancing our core production facilities and making ongoing investments in engineering, research, and development. We are also evaluating our global footprint to optimize production capabilities and meet the growing demand. We are not just focused on the immediate needs of today; we are preparing for the demands that will arise in the next six months, a year, or even two to three years. In this race for capacity, the winner will be the one best positioned with the most scalable and resilient capacity.

What is the future of cooling systems in relation to AI chips, and where do you see this race heading?

While we are not completely moving away from air cooling as it will always play a role in the equation, fully eliminating it would be prohibitively costly. The transition to liquid cooling is a critical step forward. We are already seeing advancements in the liquids used to cool servers, enabling them to absorb higher levels of heat. However, the primary challenge will be addressing the overall densification of data center systems, as more powerful and compact solutions require innovative approaches to heat management.

Our partnerships with industry leaders like NVIDIA and Intel are essential, as they provide us with invaluable, first-hand insights into the development of cutting-edge chips and GPUs. While cooling and power systems might seem straightforward, AI introduces a new layer of complexity that demands forward-thinking solutions. To meet these challenges, we are making significant investments in research and development to support the AI-driven data centers of today and tomorrow. Our commitment to continuous innovation ensures that we remain at the forefront of these critical advancements.