Lin Yifu Says Open-Source Large Models Could Help Developing Countries Narrow Income Gap

At the Intel Technology Innovation and Industry Ecosystem Conference on September 22, 2026, Justin Yifu Lin, dean of Peking University's Institute of New Structural Economics, said that core AI large models can be accessed free or at low cost through open-source approaches, giving developing countries a latecomer advantage in applying AI in manufacturing. He argued that the fourth industrial revolution could narrow rather than widen income gaps between countries.

Lin said the previous three industrial revolutions allowed pioneering countries to control key technologies, making catch-up costly for latecomers and widening income gaps. He cited research indicating that the per-capita income gap between the richest and poorest regions widened from about five times in the early 18th century to 10, 20, or even 100 times today. The open-source nature of large models, he said, changes the pace of technology diffusion.

On AI industrialization, Lin said China and the United States are both in the leading tier. The U.S., with abundant capital and hardware strengths, has developed capital-intensive, closed-source models powered by massive compute. China, by contrast, has pursued a lighter-asset path focused on open-source weights and algorithm optimization, leveraging its engineering talent and large market. Lin noted that eight of the world's ten most-used large models currently come from China.

He compared large models to the steam engine of the first industrial revolution, saying open-source AI holds similar fundamental value. He predicted that AI-plus, or the intelligent transformation of industries, would bring explosive growth and become the largest source of demand for AI industrialization.