Whalesbook
AI's Uneven Diffusion Creates Global Chasm
Microsoft vice chair Brad Smith delivered a sobering assessment at the World Economic Forum, warning that artificial intelligence's productivity boom is creating a starkly uneven playing field. His remarks, drawn from Microsoft’s AI Diffusion Report 2026, revealed a significant digital divide. While 25% of the population in advanced economies of the Global North actively uses generative AI, adoption barely reaches 14% in the Global South.
Historical Parallels and Future Infrastructure Needs
Smith drew a chilling parallel to colonial history, likening the current situation to past resource extraction in regions like India and Africa. He argued that in both eras, infrastructure development primarily served external interests rather than local populations. He stressed that failing to invest in the "infrastructure of the future"—specifically data centers and reliable electricity—in the Global South risks AI exacerbating, rather than closing, historical inequalities.
Developed World Paradox: Resistance to Progress
Paradoxically, even in developed nations, the very infrastructure required for AI expansion faces local opposition. Smith cited the United States, where concerns over water usage and job security led to the blocking of approximately $98 billion in private sector data center investments in a single quarter last year. He emphasized that for AI to be politically sustainable, tech firms must demonstrate tangible community benefits, such as reduced energy costs and improved water resources, rather than solely serving global algorithms.
Access Remains the True Battleground
International Monetary Fund managing director Kristalina Georgieva supported this view, describing the current scenario as an "accordion of opportunities." She warned that the gap between those prepared for AI and those left behind could stretch to a breaking point. The fundamental challenge, she concluded, is not the race for AI supremacy but ensuring equitable access to its transformative potential.