We keep a very clear and transparent communication about all Scala milestones. Our stakeholders are always updated and informed on our results and growth plan​




Discover Scala’s insights and innovations – your source to stay updated on trends in the industry, future-proof data centers, innovative methodologies, and sustainable practices.

Navigating the AI Revolution: The New Era of Data Center Colocation

 The growing demand for Artificial Intelligence (AI) data centers has been reshaping the landscape of global colocation services. As AI continues to advance, its needs are significantly influencing the design, construction, operation, and sustainability of those buildings. This blog post delves into the requirements for building AI infrastructure data centers, the challenges of sustaining high demand, and the overall impact on the colocation market.  

The AI Data Center Demand: A Surge in Power and Design Innovations  

AI technologies have brought forth unprecedented levels of power consumption and cooling requirements in data centers. According to Data Bridge Market Research, the AI infrastructure market is expected to reach US$422.5 billion by 2029, a testament to the rapid growth and investment in this sector. Nvidia’s CEO, Jensen Huang, predicts an even more significant expenditure, estimating that $1 trillion will be spent on data center upgrades in the next four years, primarily by Hyperscalers like Amazon, Google, Microsoft, and Meta. Scala Data Centers is at the forefront of this transformation, offering cutting-edge solutions to meet these evolving demands.  

Redefining Power Dynamics  

The traditional 5-7kW power per data center rack is no longer sufficient. The advent of AI-specific chips and servers necessitates power-dense racks, with some Graphics Processing Unit (GPU) nodes requiring as much as 10kW per unit. We are now seeing the construction of 30kW, 50kW to 100kW racks – and even higher – for AI workloads.  

Cooling: The Liquid Solution  

Traditional air-cooling technologies are not enough for the power densities required by AI applications. This shift has led to the adoption of liquid cooling solutions. Scala Data Centers is investing in such technologies for its AI & Machine Learning (ML) data centers.  

Sustainability: The Green Challenge  

As AI data centers consume more power, the focus on energy-efficient operations intensifies. According to the International Energy Agency (IEA), data centers already account for about 1.5% of global energy consumption. As of 2020, data centers and data transmission networks were responsible for approximately 330 Mt CO2 equivalent, representing about 0.9% of energy-related greenhouse gas emissions, or 0.6% of total greenhouse gas emissions globally. Looking ahead, the IEA’s Net Zero Scenario stipulates that emissions from data centers need to be halved by 2030 to align with broader environmental goals which underscores the urgent need for the data center industry to adopt more sustainable practices and technologies. Those measures are focused to meet the growing demands of AI and other advanced computing needs while mitigating their environmental impact. Scala Data Centers has been committed to sustainable practices from day one, boasting the lowest design Power Usage Effectiveness (PUE <1.3-1.4) in Latin America and implementing water-free cooling systems in its new data centers, resulting in a Water Usage Effectiveness (WUE) = ZERO.  

The Rise of Edge Computing  

The rise of AI has not only transformed various industries but has also accelerated the significance of edge computing. The demand for instantaneous decision-making, particularly in applications like chatbots, has underscored the necessity for inferencing using AI models to operate near the data source. This demand extends beyond the gaming industry, prompting innovation in the realm of edge data centers. Scala Data Centers is at the forefront of this innovation with its proprietary design and construction methodology known as FastDeploy. The solution created by the company’s Center of Excellence in Engineering (CoE), is a highly replicable and scalable solution based on tailor-made prefabricated and transportable modular components that allow our Hyperscale customers to enable new regional markets in record time. By utilizing a modular approach, a single shell is built, with components incorporated upon demand, offering multiple advantages for HyperEdge buildings:  

💡 Unprecedented expansion capacity: independent capacity blocks added upon demand allow for long-term contracts with smooth ramps of growth  

💡 Future-proof: we can benefit from new technologies that come along  

💡 Reduced carbon footprint: with fewer stranded assets, carbon footprint is reduced  

Although using modular components, the Hyperscale customers have the same high-quality experience as traditional data centers.  


The LATAM Perspective: A Growing Market with Unique Challenges  

In the LATAM region, the growth of AI data centers presents both opportunities and challenges. The region is experiencing a surge in demand for data center services, driven by digital transformation and the expansion of cloud services. However, infrastructure challenges, such as uneven clean power availability and different temperatures across the region, pose significant hurdles.  

Scala Data Centers is investing heavily in the region, building data centers that cater to the high-power and cooling demands of AI workloads. These facilities are designed to be scalable and flexible, sustainable and future-proof, allowing for rapid expansion as AI requirements evolve.  


Scala Data Centers – Shaping a Future-Ready Industry  

The AI revolution in data centers is a dynamic journey. As AI evolves, so do the requirements for data center design, operation, and sustainability. Colocation providers must innovate continuously to meet the growing demands of AI infrastructure. Scala Data Centers is not just housing data; we are enabling the AI-driven technologies reshaping our world, positioning ourselves as a leader in the future-ready data center industry.  


Scala’s Blog