The global landscape of artificial intelligence (AI) is increasingly marked by a stark divide between nations with advanced computing power and those left behind. In May, Sam Altman, CEO of OpenAI, visited a massive new data centre project in Texas, estimated to cost $60 billion and surpass the size of New York’s Central Park. Scheduled for completion next year, this centre is set to become one of the most powerful computing hubs globally. Meanwhile, in Argentina, Nicolas Wolovick, a computer science professor at the National University of Cordoba, operates one of the country’s most advanced AI facilities from a converted room filled with outdated technology. Wolovick expressed concerns about the growing gap, stating, “Everything is becoming more split. We are losing.”
AI technology has created a new digital divide, reshaping geopolitics and global economics. The disparity in computing power influences dependencies among nations and ignites a rush to participate in a technology race. The United States, China, and the European Union stand as the primary beneficiaries, hosting over half of the world’s most powerful data centres, as reported by researchers at Oxford University. Only 32 countries, roughly 16 percent of the global total, have large-scale facilities capable of supporting cutting-edge AI systems.
The data indicates that the US and China dominate the technology sector, operating more than 90 percent of the data centres utilized for AI development. In stark contrast, Africa and South America lack significant AI computing hubs. Countries like India and Japan possess a minimal number of facilities, while over 150 nations have none.
Today’s AI data centres require substantial investment and infrastructure, often costing billions to build. The concentration of ownership among a few tech giants exacerbates the divide. AI systems, particularly those powering popular applications like ChatGPT, perform better in languages prevalent in regions with concentrated computing power, such as English and Chinese. Advanced computing capabilities are essential for breakthroughs in fields such as drug discovery, gene editing, and even military applications.
Nations without adequate AI infrastructure face significant challenges. These include limits on scientific research, obstacles for emerging companies, and difficulties in retaining talent. Officials have expressed alarm over the reliance on foreign corporations for critical computing resources. As Vili Lehdonvirta, an Oxford professor involved in the research, noted, “Oil-producing countries have had an oversized influence on international affairs; in an AI-powered near future, compute producers could have something similar.”
The significance of AI computing power has made components like graphics processing units (GPUs) pivotal to trade policies, especially for the US and China. As nations scramble to develop their AI infrastructure, some are allocating public funds to enhance their technological autonomy. The Oxford research team methodically mapped AI data centres, uncovering trends that highlight the concentration of resources among a few major companies.
According to the research, US companies operate 87 AI computing hubs, constituting nearly two-thirds of the total global facilities. In comparison, Chinese firms manage 39, while European companies oversee just six. The majority of chips used in these centres are sourced from prominent manufacturers like Nvidia.
“We have a computing divide at the heart of the AI revolution,” stated Lacina Kone, director general of Smart Africa. He emphasized that this divide extends beyond hardware to issues of digital sovereignty.
Despite a decade of progress in reducing the tech gap through increased internet access and affordable smartphones, the UN recently warned that the digital divide could widen without proactive measures. The agency highlighted that just 100 companies, primarily from the US and China, accounted for 40 percent of global AI investment, giving them significant control over the technology’s future.
The scarcity of GPUs, essential for AI development, stems from the need for multibillion-dollar manufacturing facilities. As demand surges, prices for these chips have escalated, creating a competitive landscape where only a few countries can secure them.
Countries lacking AI infrastructure often resort to renting computing power from distant data centres, facing issues like high costs and slower connection speeds while being subject to the regulatory environments of foreign companies. Brad Smith, president of Microsoft, confirmed that many nations desire more local computing facilities for greater sovereignty. However, achieving this in regions like Africa, where reliable electricity is often a challenge, remains daunting. Microsoft is actively working on building a data centre in Kenya in partnership with the UAE-based company G42, focusing on market demand and local resources.
In response to the growing imbalance in AI capabilities, various countries are implementing strategies to close the gap. They are offering incentives for land and energy resources, expediting permit processes, and investing in chip acquisition and data centre construction. For instance, the government of India is subsidizing AI infrastructure to foster local language models, while several African nations are discussing the establishment of regional computing hubs. Brazil has committed $4 billion to AI initiatives, and in February, the European Union announced plans to invest €200 billion in AI projects, including new data centres.
Despite these efforts, bridging the AI divide will likely require collaboration with major powers such as the United States or China. The future landscape of AI will not only reshape industries but could redefine global power dynamics as nations vie for technological supremacy.
