April 2026. In the first quarter of 2026, open-weight AI models crossed a threshold. DeepSeek trained a frontier model for $5.6 million. A Bulgarian physicist built the most widely used inference engine in the world from his laptop. India released its first sovereign language models. France produced three AI billionaires. Meta abandoned open release. NVIDIA committed tens of billions to fill the vacuum. A hedge fund in Beijing published the world's leading coding model on New Year's Day. Yann LeCun walked out of the company he had served for twelve years. And forty-one percent of all model downloads on the world's largest AI platform now come from China. Four acts. Seven countries. The people who chose to give intelligence away, and what happens now that they have. ~38,000 words.

This episode was produced using the Proxima.Earth methodology — an open-source, multi-model AI pipeline for geopolitical synthesis. No human is in the loop after subject selection. The methodology is the editorial control. Full methodology, prompts, and production transparency: proxima.earth/methodology

Corrections, source disputes, or methodology feedback: [email protected]