What ChatGPT Really Needs to Work — And Why Infrastructure Matters

My name is Jeffrey Mdala, an AI Engineer and Founder of Zambian Online Education Company (ZOEC) in Lusaka, Zambia. I build AI-powered education tools like eskulu, a learning platform designed around the Zambian ECZ curriculum, and I have spent years thinking deeply about what it really takes to make AI useful in African contexts.

One thing many people see is the magic of tools like ChatGPT. You type a question, and within seconds you get an answer, an explanation, a summary, or even a draft of an idea. But behind that simplicity is a much bigger story: AI does not run on magic. It runs on infrastructure, computing power, data systems, and enormous energy demands.

That is the part of the conversation I believe we need to understand more clearly in Zambia and across Africa.

AI feels simple on the surface, but it is built on massive systems

When most people interact with ChatGPT, they experience it through a phone or laptop. It feels lightweight. It feels instant. It feels like something happening inside an app. But the reality is very different.

Systems like ChatGPT depend on powerful data centres, high-performance computing hardware, large-scale storage, networking infrastructure, and continuous optimization. These systems process huge volumes of requests from users around the world. That means every prompt, every response, and every generated output depends on machines running at scale somewhere in the background.

In simple terms, AI tools may look small on your screen, but behind them are industrial-level systems.

Why data centres matter in the AI era

Data centres are becoming one of the most important foundations of the modern digital economy. They host the servers that power cloud platforms, websites, business systems, and increasingly, AI applications.

For generative AI in particular, the demand is even higher because these models require:

• Heavy computational power to process prompts and generate responses
• Large memory and storage systems to manage model operations and supporting services
• Reliable internet infrastructure for fast access across regions
• Stable electricity to keep systems available at all times
• Cooling systems to manage the heat produced by intensive computing

This is why conversations about AI should never be separated from conversations about infrastructure. If Africa wants to participate meaningfully in the AI economy, we must think beyond just using AI tools. We must also think about the systems that make them possible.

The hidden cost of AI is not just software

One of the biggest misconceptions about AI is that it is mainly a software story. In reality, it is also a hardware, energy, and logistics story.

Running advanced AI systems requires expensive chips, strong cloud infrastructure, and significant energy consumption. As demand for AI grows, so does the pressure on data centres and power systems. This is one reason why global technology companies continue investing heavily in infrastructure instead of only building apps.

For African innovators, this matters because it shapes what is practical, affordable, and scalable. It affects how we deploy AI products, where we host them, how fast they respond, and how much they cost to maintain.

As someone who has built digital education platforms used by hundreds of thousands of learners, I have seen firsthand that technology success is not just about writing code. It is about designing for real-world constraints: bandwidth, affordability, device limitations, hosting costs, and user behavior in our local environment.

What this means for Zambia and Africa

In Zambia, and across much of Africa, the AI conversation is often framed around adoption: how to use ChatGPT, how to automate tasks, how to improve education, how to support business productivity. Those are important questions. But we should also ask deeper ones:

• Where will the infrastructure for African AI growth come from?
• How do we build systems that work well under local constraints?
• How do we create AI solutions that are not just imported, but adapted to our realities?
• How do we prepare students and businesses for an economy shaped by intelligent systems?

I believe this is where African founders, engineers, policymakers, and educators need to work together. We need local innovation, yes, but we also need long-term thinking around digital infrastructure, cloud systems, education, and policy.

That is part of why I built eskulu. My goal was never just to put content online. I wanted to create a platform that could intelligently support Zambian learners with notes, past papers, quizzes, marking schemes, and AI-powered assistance in a way that fits our education system. Today, eskulu has reached 500,000+ students across Zambia, and that impact has only strengthened my belief that African problems need African-built systems.

Building AI in Africa requires practical thinking

My own journey into technology started long before AI became a global trend. I started coding in Grade 12 after graduating as the best student at Thornhill Boarding and Day School. Since then, I have kept learning across computing, engineering, AI, mathematics, and digital systems.

I have built products, worked in technical roles, and continued studying while developing real solutions for local users. Along the way, I was recognized in the Top 5 of the ZICTA Innovation Programme for eskulu, won Business With a Purpose at the X Pitchathon by Accessbank and MTN in 2023, and earned 3rd place at the Yango and Zindi Data Science Hackathon in 2024.

These milestones matter to me not just as awards, but as proof that world-class innovation can come from Zambia when we focus on real needs and execute with discipline.

In Africa, practical AI matters more than hype. We need solutions that work on everyday phones, in schools with limited resources, in businesses that cannot afford waste, and in environments where infrastructure still has gaps. The future will belong to builders who understand both the promise of AI and the reality of deployment.

AI opportunity is real, but so are the challenges

I am optimistic about AI because I have seen how it can expand access to knowledge, improve productivity, and unlock new business models. In education especially, the potential is enormous. A well-designed AI system can help a student revise more effectively, understand difficult concepts faster, and access learning support at any time.

But optimism must be matched with honesty. AI systems are resource-intensive. They depend on infrastructure that many African countries are still developing. They require thoughtful implementation, not blind excitement.

That is why I believe our role as African technologists is not simply to repeat global narratives. It is to interpret them for our context. We must build with awareness of our constraints, our opportunities, and our long-term goals.

My perspective as a founder and AI engineer

As the founder of ZOEC and someone who has also worked as an AI Engineer, including at Unicaf University Zambia, I see AI as both a technical field and a nation-building opportunity. It is not just about models and prompts. It is about education, productivity, access, and economic participation.

It is also about preparing the next generation of Zambian and African builders to understand what sits behind the interface: the servers, the energy, the networks, the cloud architecture, and the engineering decisions that turn an idea into a reliable system.

That understanding is what will help us move from being consumers of AI to becoming serious contributors to its future.

Conclusion

ChatGPT and similar tools may feel effortless, but they depend on a deep and expensive infrastructure stack. That reality should not discourage us. Instead, it should sharpen our thinking.

For Zambia and Africa, the AI future will not be built by hype alone. It will be built through infrastructure, education, local problem-solving, and founders willing to create tools that serve real people in real conditions.

That is the work I care about most. Through eskulu, through ZOEC, and through my AI consulting work, I remain committed to building technology that is useful, grounded, and impactful for our region.

If you are interested in AI-powered education, digital product development, or AI consulting for your organization, feel free to reach out to me at jeffmdala@gmail.com. You can also explore the vision behind eskulu as we work toward a future where every learner in Zambia has access to intelligent academic support.