AI computing divide leaves 150+ nations without critical infrastructure

Artificial intelligence computing power is creating a stark global divide, with only 32 countries hosting AI-specialized data centers while more than 150 nations have no such infrastructure. This digital gap is reshaping geopolitics and economics, as nations with advanced AI capabilities gain significant advantages in scientific research, business automation, and technological sovereignty, while those without face mounting challenges in talent retention and economic development.

The big picture: The United States, China, and the European Union dominate the AI computing landscape, hosting more than half of the world’s most powerful data centers used for developing complex AI systems.

• American companies operate 87 AI computing hubs globally, representing nearly two-thirds of the total, compared to 39 operated by Chinese firms and six by Europeans, according to Oxford University research.
• More than 90 percent of data centers used by other companies for AI work are operated by American and Chinese firms.
• Africa and South America have almost no AI computing infrastructure, while India has at least five facilities and Japan has at least four.

Why this matters: The concentration of AI computing power is creating new forms of technological dependency and influencing global power dynamics in ways reminiscent of oil-producing nations.

• “Oil-producing countries have had an oversized influence on international affairs; in an A.I.-powered near future, compute producers could have something similar since they control access to a critical resource,” said Vili Lehdonvirta, an Oxford professor who conducted the research.
• Nations without AI compute power face limitations in scientific breakthroughs, including drug discovery and gene editing, while also struggling with brain drain as talent migrates to countries with better resources.

What’s driving the divide: The global shortage and high cost of specialized microchips called graphics processing units (GPUs) are creating bottlenecks for AI development.

• GPUs, mostly manufactured by Nvidia, a leading semiconductor company, require multibillion-dollar factories to produce and are packed into data centers by the thousands to provide computing power for AI models.
• These data centers cost billions to build and require massive amounts of electricity and water, infrastructure that many countries cannot provide.
• The world’s most advanced AI systems are more proficient in English and Chinese—languages of countries where compute power is concentrated.

Real-world impact: The computing divide is already affecting businesses and researchers in developing nations who must work around infrastructure limitations.

• Qhala, a Kenyan startup building an AI system for African languages, rents computing power from data centers outside Africa and schedules work during American nighttime hours to avoid network congestion.
• In Argentina, computer science professor Nicolás Wolovick operates one of his country’s most advanced AI hubs in a converted university room with aging chips, watching his top students leave for the United States and Europe where they can access GPUs.

Geopolitical implications: The United States and China are using their AI computing advantages to exert global influence through trade restrictions and strategic partnerships.

• The Biden and Trump administrations have controlled which countries can buy powerful AI chips, while China has used state-backed loans to encourage sales of its companies’ networking equipment.
• In the Middle East, an Emirati company promised to exclude Chinese technology in exchange for access to AI technology from Nvidia and Microsoft, while President Trump signed additional deals giving Saudi Arabia and the UAE greater access to American chips.

What countries are doing: Alarmed by the concentration of AI power, many nations are investing public funds to build “sovereign AI” capabilities.

• Brazil has pledged $4 billion for AI projects, with President Luiz Inácio Lula da Silva saying: “Instead of waiting for A.I. to come from China, the U.S., South Korea, Japan, why not have our own?”
• The European Union outlined plans to invest 200 billion euros for AI projects, including new data centers across the 27-nation bloc.
• India is subsidizing compute power and creating AI models proficient in the country’s languages, while African governments are discussing regional compute hubs.

Key stats: The scale of AI investment highlights the magnitude of the global divide.

• American tech giants Amazon, Microsoft, Google, Meta, and OpenAI have pledged to spend more than $300 billion this year on AI infrastructure—approaching Canada’s national budget.
• Harvard’s Kempner Institute has more computing power than all African-owned facilities on that continent combined.
• Cassava, a tech company building Africa’s most advanced data centers, expects its $500 million, five-facility project to address only 10-20 percent of the region’s AI demand.

What they’re saying: Industry leaders acknowledge the challenges while working on solutions.

• “The A.I. era runs the risk of leaving Africa even further behind,” said Brad Smith, Microsoft’s president.
• “Africa will strike a deal with whoever can give access to GPUs,” said Lacina Koné, director general of Smart Africa, an organization that coordinates digital policy across the continent.
• “I don’t think Africa can afford to outsource this A.I. sovereignty to others,” said Hardy Pemhiwa, Cassava’s chief executive.