AI data centers don't have to be isolated energy drains. Two ambitious projects on opposite sides of the Atlantic are demonstrating that high-performance computing infrastructure can be designed to serve regional economies and environmental goals simultaneously. Mexico is building Latin America's first sovereign AI cloud platform, while Denmark has just launched a supercomputer that feeds its waste heat directly into the local energy grid.

Why Are Countries Building Their Own AI Infrastructure Instead of Relying on U.S. Tech Giants?

Digital sovereignty has become a critical concern for governments and enterprises worldwide. Mexico's AI Green Data Centers (AI-GDC) is positioning itself as an alternative to U.S.-dominated cloud providers by keeping data, computations, and AI models within Latin American borders. This matters because it ensures compliance with local regulations and protects sensitive information from foreign access.

AI-GDC, headquartered in Mexico and led by CEO Franz Berchelmann, is developing infrastructure in northern Mexico designed specifically as an NVIDIA-aligned AI Factory. The company is working toward becoming an NVIDIA Cloud Partner, validating its ability to deliver enterprise-grade GPU infrastructure at scale. Construction of its flagship green data center campus is scheduled to begin this year, with plans to scale up to 100 megawatts of capacity by 2030.

"Latin America is emerging as an important frontier for AI infrastructure investment. With billions of dollars flowing into Mexico's AI data centers, the operators who will win will move beyond commodity GPU rental and offer differentiated AI services at scale," said Haseeb Budhani, CEO and co-founder of Rafay Systems.

Haseeb Budhani, CEO and co-founder of Rafay Systems

The market opportunity is substantial. Global AI infrastructure spending is forecast to reach $758 billion by 2029, and within Mexico specifically, the AI data center market is projected to grow at a 24.55 percent compound annual growth rate through 2031.

How Can Data Centers Reduce Their Environmental Impact While Supporting AI Workloads?

• Waste Heat Recovery: Denmark's new supercomputer, named Bitten, uses advanced liquid cooling with full heat recovery. The waste heat is reused as part of Sønderborg Municipality's goal to create a fully carbon dioxide neutral energy system, turning what would normally be an environmental cost into a community resource.
• Water-Efficient Cooling Systems: AI-GDC is incorporating water-efficient cooling systems designed specifically for the high-density power and cooling requirements of next-generation GPUs, minimizing environmental footprint while maintaining performance.
• Renewable Energy Integration: Both projects prioritize renewable energy sources as part of their infrastructure design, reducing reliance on fossil fuels and aligning with regional climate goals.

The University of Southern Denmark, in collaboration with Danfoss and Hewlett Packard Enterprise (HPE), has created a model that proves sustainability and technological advancement can reinforce each other. Bitten will serve researchers and students across Danish universities, enabling work on larger datasets and more complex artificial intelligence models within a single shared infrastructure.

"This new system is among the most advanced in Denmark for generative AI workloads and represents a significant technological upgrade of the national AI infrastructure," explained Carsten Nielsen, Vice President and Managing Director for the Nordic Cluster at HPE.

Carsten Nielsen, Vice President and Managing Director for the Nordic Cluster at HPE

What Services Will These New Data Centers Actually Provide?

AI-GDC is moving beyond traditional GPU-as-a-service offerings to deliver a full spectrum of AI capabilities, from bare metal infrastructure to complete AI models. The company is building an application marketplace tailored specifically for regional needs, targeting government and public sector organizations, enterprises in manufacturing and logistics, and leading research institutions like Tecnológico de Monterrey.

The partnership with Rafay Systems enables AI-GDC to provide self-service, token-metered AI use cases. This means customers can access computing resources on demand, paying only for what they use, similar to how cloud storage works. The orchestration layer that Rafay provides helps maximize infrastructure utilization while maintaining security and compliance with regional regulations.

Denmark's Bitten supercomputer is being made available through UCloud, a European research cloud platform developed collaboratively by the University of Southern Denmark, Aalborg University, and Aarhus University. UCloud currently has more than 23,000 users and is among the largest research cloud platforms in Europe. This ensures that data, software, and computations remain under national and European control, addressing growing concerns about digital sovereignty.

"Researchers and students now have much better opportunities to work with larger datasets and more advanced models across institutions than has previously been practically possible in a shared Danish infrastructure," stated Professor Claudio Pica, Director of the SDU eScience Center.

Professor Claudio Pica, Director of the SDU eScience Center

Both projects signal a broader shift in how AI infrastructure is being developed globally. Rather than concentrating computing power in a handful of U.S. data centers, regions are investing in localized, sovereign alternatives that serve their specific economic and regulatory needs while demonstrating that environmental responsibility and cutting-edge AI capability are not mutually exclusive.