Generative artificial intelligence (GenAI) is no longer just a technology of the future; it has become a transformative application. Unlike traditional AI, which involves performing specific tasks using defined algorithms and rules, GenAI creates new content from large volumes of data. But it is much more than content generation: it fuels innovation by enabling real-time decisions, streamlining complex processes, and accelerating breakthroughs. This rapid growth also creates unprecedented compute demand and higher rack densities, requiring new generations of advanced cooling, resilient power, and scalable architectures to sustain AI workloads.
The path forward demands concrete answers: How can data centers be adapted to support the increased data processing, temperatures, and energy consumption that AI requires?
While these challenges are global, Latin America is experiencing them in its own way. Digital transformation was already underway in the region, but the pandemic accelerated adoption. In markets like Mexico, companies are not only adopting GenAI but also monetizing it, leveraging AI to reach new markets, reduce costs, and improve customer experiences.
More AI means more heat
More processing means more heat generation. Currently, there are racks that consume over 100 kW, and projections show this could reach up to 1 MW in the coming years. This requires more energy not only to operate, but also to cool this mission-critical equipment. Traditional air cooling is no longer sufficient for new, high-density GPUs. These systems require advanced cooling solutions, such as liquid cooling or hybrid air-liquid deployments, which remove heat both from the data center environment (air) and directly from the GPUs in the racks (liquid). As a result, it's not just about higher cooling costs—new technologies are essential, and liquid cooling is inherently more energy efficient compared to air cooling alone:
- Direct-to-chip liquid cooling can efficiently eliminate approximately 70-75% of the heat generated by the equipment in a rack, leaving 25-30% of the heat for air cooling systems to remove in the room.
- Immersion cooling systems immerse servers and other components in a thermally conductive dielectric fluid or liquid, eliminating the need for air cooling. This approach maximizes the liquid's thermal transfer properties and is the most energy-efficient form of liquid cooling on the market.
- Rear door heat exchangers replace the rear door of the IT equipment rack with a liquid heat exchanger that can be used in conjunction with air cooling systems to cool environments with mixed rack densities.
Using a liquid cooling strategy removes heat at the source and can reduce data center PUE metrics. Some liquid cooling systems also allow equipment to recover and reuse heat. These gains can reduce indirect or energy-regulated emissions for businesses. As a result, liquid cooling can be an essential part of companies' sustainability programs.
In the video podcast What Is Next? with Vertiv, Amet Novillo, president of Equinix Mexico and the Mexican Data Center Association, discusses several initiatives, including Equinix Heat Export. This system extracts the heat from chips and servers and delivers it to a heat provider for reuse. With rising energy prices and ongoing global disruption, access to this heat source mitigates price volatility and energy insecurity for local authorities. Equinix has set a corporate goal to use 100% renewable energy by 2030, investing in efficiency and innovation programs that address both energy use and water consumption.
Data center modernization for the AI era
The evolution of critical infrastructure requires rethinking the design of data centers, whether renovating an existing one or constructing a new one. Greenfield projects start on undeveloped land and have the advantage of strategizing optimized designs, full customization, and scalability for future technologies. Brownfield projects focus on upgrading existing facilities to support high-density and mixed loads - enabling lower acquisition costs, faster implementation, and the advantage of established locations.
Regardless of the strategy, operators must also consider everything from the physical weight of the new racks and equipment, to selecting systems that allow scalable growth for faster deployment. But modernization is not just about facilities and equipment, it also depends on people. As Amet Novillo highlighted, building and operating AI-ready facilities requires skilled professionals in power, cooling, and IT management skills that are still scarce in the region. This also extends to more specialized areas of the sector, such as structured cabling, fiber optics, and other sensitive systems.
Looking ahead, the impact of AI will also be defined by the industries that embrace it first. For example, in Latin America, Mexico is already seeing accelerated adoption in sectors such as retail, financial services, manufacturing, and telecommunications, where AI is being applied to improve supply chains, enhance customer experiences, and increase operational efficiency.
Vertiv’s comprehensive and growing portfolio of critical digital infrastructure solutions is designed to support the growth and development of the digital transformation initiatives and AI ecosystems. With its teams and talent, Vertiv is ready to partner with LATAM customers and support business strategies to harness the full potential of AI.
