Artificial intelligence (AI) has rapidly transformed from an experimental concept into the foundation of critical infrastructure worldwide, generating new demands on data centers. Following the launch of ChatGPT in November 2022, the discussion around AI accelerated sharply, evolving from initial curiosity into a strategic priority for organizations across various industries.
Dion Chang, founder of Flux Trends and a trend analyst, described the pace of these changes at a joint event organized by Vertiv and its client Open Access Data Centres (OADC). He noted that over three years, the landscape has undergone colossal transformations: generative AI was initially used for simple tasks like writing emails and summarizing documents, and subsequently, agentic AI systems have emerged, which begin to manage tasks autonomously.
Wojtek Piorko, Managing Director of Vertiv for Africa, explained this transformation by pointing to the influence of Graphics Processing Units (GPUs). He compared the Central Processing Unit (CPU) to a single-lane road where vehicles move sequentially, whereas the GPU is like a multi-lane highway allowing thousands of vehicles to move in parallel. This capability to perform thousands of simultaneous calculations is critically important for AI.
As a result of this process, there is a significant leap in equipment density. What previously required several racks can now fit into one. While this increases space utilization efficiency, it fundamentally changes the requirements for cooling and power supply systems, as well as the overall design of data centers.
With the advent of AI, rack density is shifting from 30–40 kW to potentially hundreds of kilowatts in the coming years, making support—energy, cooling, and fiber optics—more crucial than ever. Piorko emphasized that AI deployments raise serious questions, primarily concerning electricity availability, as AI-based data centers require hundreds of megawatts of additional power. He added that although power generation in Africa has roughly doubled over the last decade, this may not be enough, and the main issue now is energy distribution, which is particularly relevant for South Africa.
According to Piorko, currently, less than half a gigawatt of active capacity is available in Africa for a population exceeding one billion people, meaning Dubai has the same capacity as the entire continent. He concluded that achieving catch-up requires stable power supply, regulatory compliance, connectivity, and a skilled workforce.
Mark Matthews, Engineering Director and Project Manager at OADC, stated that achieving these goals is possible thanks to the company's own successes on the continent. He noted that the OADC model, operating as a cable landing station in Amazinmotohti, South Africa, demonstrates that data centers are useless without connectivity. Without access to subsea cables and global networks, data cannot reach users.
One of the major shifts in the local industry is the transition to edge computing. Matthews explained that if a user in Johannesburg requests restaurant recommendations from an AI system, they should not have to wait a minute for a response from a data center on another continent; the AI output must occur closer to the user.
In terms of sustainability, Matthews mentioned that in Nigeria, OADC is studying gas generation as an alternative to diesel fuel, and in South Africa, solar panels are used to offset the energy consumption of one data hall during the day. He also noted that small operational adjustments, such as changing cooling settings, can significantly save energy.
Regarding water resources, Matthews clarified that modern cooling systems operate in closed loops, allowing the same water to be reused multiple times. Piorko agreed, mentioning the recently launched Vertiv SmartRun system, which integrates high-density power distribution, closed-loop liquid cooling, network infrastructure, and isolation infrastructure into a single modular platform. This design ensures a faster and more efficient deployment process.
Modularity is part of the company's overall philosophy: components are manufactured and tested in factories before being shipped to the site, which reduces risks, accelerates deployment, and minimizes construction waste. This approach was implemented at the OADC facility in Lagos, Nigeria, which was initially planned for 24 MW but was launched with 6 MW with the possibility of subsequent scaling up to 24 MW to support both current and future AI needs.
These approaches indicate the formation of a more scalable and sustainable model for data center development in Africa, balancing current demand with how AI will impact infrastructure in the coming years.
Artificial intelligence (AI) is often viewed in extreme terms: either as a threat that will replace workers or as a tool that facilitates the next wave of growth and productivity. However, the actual picture is likely to be more complex. The potential impact in South Africa will probably differ from what is observed in economies like the US, although developments there may offer some clues about what to expect locally.
Studies by the International Monetary Fund show that about 40% of global employment is somehow affected by AI. In developed countries, this figure rises to approximately 60%. Goldman Sachs predicts that over the next decade, AI will affect around 300 million jobs worldwide.
Interestingly, we are not currently seeing mass job losses. Instead, AI appears to be transforming how people enter the labor market and advance their careers. The greatest pressure is felt in the service sector and at entry-level positions, where companies may be more cautious about creating new roles than before.
The impact also depends on the composition of the workforce. Positions that are highly administrative, repetitive, routine, and process-based are most at risk. Conversely, higher-level positions seem more protected. These include jobs where experience, judgment, and oversight are important, especially those related to evaluating AI-generated results and determining their accuracy and relevance.
There is also the argument that while AI may eliminate some jobs, the technology will likely create new ones. Roles such as programmers, engineers, and process developers may benefit from these changes.
In the US, graduate absorption has slowed down as companies have been less active in creating new vacancies than before. A study by IESE Business School showed that the wages of junior employees in companies affected by AI decreased after the launch of ChatGPT. Even for those who find jobs, wage growth has been unfavorable.
In a country like South Africa, where the economy already faces a significant youth unemployment problem, many graduates are already struggling to find meaningful first opportunities. If AI reduces the need for entry-level positions or slows their creation, it could further complicate the transition from education to work.
This is both a social and a business issue. Many organizations naturally invest in technology to improve margins, but they also need to look beyond short-term efficiency. Succession planning, knowledge transfer, and capacity building remain critical. If businesses stop attracting and developing young talent internally, they risk weakening the very pipeline that will provide future leaders and oversee increasingly complex systems.
The same applies to how AI is beginning to merge with robotics. In manufacturing, the future is not just robotics, but the integration of robotics and AI. This shift may increase the demand for highly skilled technical oversight while reducing the number of routine positions. For a country already grappling with inequality, the issue shifts from jobs displaced by technology to how workers prepare and equip themselves for the roles that technology creates.
For workers, AI technical literacy is important, but human capabilities that are harder to automate—including judgment, relationship building, accountability, and the ability to interpret results in context—are equally vital.
From an investment perspective, it is important to seek out companies that do not focus solely on margin expansion. There is a natural limit to cost savings, whereas growth is less constrained. AI strategies should increasingly be evaluated based on their contribution to revenue growth. The winners will not be the companies that automate the most, but those that use AI for growth while maintaining judgment, accountability, and the talent pipeline that sustains the business over time.
