AI Ambitions Face Power And Cooling Challenges

While the world is celebrating ChatGPT and all the GenAI breakthroughs, a different kind of revolution is happening in Africa’s server rooms – the unglamorous work of keeping machines humming in 40°C heat. AI ambitions are soaring. From Lagos to Nairobi, with start-ups building everything from agricultural AI to financial inclusion platforms. Countries are announcing national AI strategies. Tech hubs are growing. But underneath this digital euphoria lies a sobering reality: most of Africa’s existing data centre infrastructure simply cannot handle the computational demands of modern AI.

AI-dedicated data centres are uniquely designed for the distinct properties of AI workloads: high computing power, a power-hungry rack running AI training that might consume as much electricity as 10 homes, all concentrated in a six-foot-tall cabinet, and the additional hardware (such as liquid cooling) that comes with it. This isn’t just about having more servers, it’s about fundamentally reimagining how data centres (will) work.

The Power Problem That Nobody Talks About

Here’s what the AI evangelists might not tell you: power demand from AI data centres could grow more than thirtyfold by 2035. For a continent where countries across the region grapple with limited energy access, high electricity costs, and outdated infrastructure punctuated with frequent outages and heavy reliance on imported fuel sources, this creates a fundamental challenge.

This is where Vertiv enters the story—not as a flashy tech company, but as the infrastructure specialist solving problems that most people never think about but that determine whether Africa’s AI dreams can become reality.

With just about everyone and their cat chasing the next AI unicorn, Vertiv is focused on a different kind of innovation: transitioning to more complex liquid- and air-cooling strategies that can handle the heat generated by AI computations. Their work is not flashy, as it involves thermal management systems, uninterruptible power supplies, and integrated infrastructure solutions. Yet without these systems, even the most brilliant AI algorithm is just code that can’t run.

This brings us to the consumption-innovation dilemma. Jon Abbott, Technologies Director & Industry Advisor, Vertiv EMEA, explains we need to “scale up before you scale out. And that’s really important. What that means is that when we talk about AI, we talk about training. Everyone’s training the model. You need an engine large enough to train the model to a certain extent, for it to become smarter or smart enough to fulfil whatever task you’re asking of it. The purpose of this training is to bring the applications running on it much closer to the market that it serves.” Abbott adds, “It’s once you scale out, which admittedly everybody needs to still wait for, that AI will truly become tailored for a specific country or market.”

Africa’s Unique Infrastructure Challenge

Our data centre landscape presents challenges that don’t exist in Silicon Valley or Singapore. The reliability of municipal power supplies is still a big challenge. The higher computing power needed, the more cooling power needed, making traditional data centre approaches inadequate for AI workloads. It means developing cooling systems that work efficiently in Lagos’s humidity, power systems that can handle Nairobi’s grid fluctuations, and solutions that can be deployed in markets where traditional data centre construction might take years.

Let’s consider the numbers: According to McKinsey & Company, generative AI alone could add the equivalent of up to $4.4 trillion annually to the global economy, including up to $100 billion in the telecommunications sector, $130 billion in media, and $460 billion in high tech. Vertiv’s whitepaper, AI Workloads And The Future Of IT Infrastructure, indicates “The inputs associated with AI are creating a potential bottleneck in chip size, rack weight, and data centre power and cooling. Spending on critical digital infrastructure for GenAI will top $18 billion in 2024, increasing to more than $48 billion by 2027.” This growth represents not just economic opportunity, but the chance to build AI infrastructure that works for African conditions from the ground up.

This does present an opportunity. The regional data centre market is projected to grow by 50 per cent in terms of capacity by 2026, according to the Data Centres in Africa Focus Report 2024 by the African Actors of Data Centre Association (ADCA), published in May 2024. This means that Africa can do what it is primed for: leapfrog to AI-ready infrastructure rather than retrofitting legacy systems. What makes Vertiv’s approach particularly relevant for Africa is its understanding that infrastructure must be designed for local conditions.

The Compound Effect Of Infrastructure Investment

The impact of robust AI infrastructure extends far beyond individual companies. When a Nigerian fintech can reliably run machine learning models for fraud detection, it increases financial inclusion. When a Kenyan agricultural start-up can process satellite imagery in real-time, it improves food security. When a South African healthcare company can run diagnostic AI without worrying about system failures, it saves lives.

Vertiv’s work creates what economists call a “multiplier effect.” Every dollar invested in reliable AI infrastructure enables multiple dollars of innovation on top of it. A stable, AI-ready data centre in Lagos doesn’t just serve one company—it becomes a platform for dozens of start-ups to build upon. This is why infrastructure companies are crucial to Africa’s AI future. They’re not building the applications that grab headlines, but they’re creating the foundation that makes those applications possible at scale.

As Africa’s AI ecosystem matures, the lesson is clear: innovation, the sustainable kind, requires infrastructure-first thinking. The future of AI in Africa won’t be defined solely by who builds the most sophisticated algorithms, but by who can pair innovation with reliable, efficient systems designed around local needs. It’s an AI revolution that has to be built from the ground up—literally. While the world focuses on generative AI breakthroughs, Africa’s real breakthrough might be the unglamorous work of building data centres that can handle AI workloads reliably, efficiently, and with sustainability practices in mind. And, it is being done by companies like Vertiv, making sure those systems work, 24/7, in the heat and humidity of the sweltering African sun.