Global AI divide: Who controls the computing power?

 Global AI divide: Who controls the computing power?

A new report reveals a growing global divide in access to AI computing power. According to Oxford University data, only 32 countries, primarily in the northern hemisphere, have the necessary data centers to develop and run generative AI systems. Leading the charge are the U.S., EU, and China, collectively hosting over half of the world’s AI infrastructure. With the U.S. boasting 26 centers, the EU 28, and China 22, this disparity creates a significant “computing divide.”

Countries lacking these resources are often forced to rent remote compute power, facing high costs and inconvenient access times. This has become a challenge for AI startups in developing nations, such as Kenya’s Qhala and Amini, which struggle to compete without affordable access to high-performance hardware like GPUs. The cost of AI hardware remains a major barrier to entry, as companies like Elon Musk’s “Colossus” data center invest in massive infrastructures to power the next generation of AI.

As geopolitical dynamics shift, computing power is increasingly becoming the new currency of global influence—reminiscent of oil’s role in the 20th century. With nations aligning either with U.S. or Chinese infrastructure, the race for AI supremacy is now about who controls the computational resources that fuel these technologies.