Patrick Foulis, an FT contributing editor and Hoover Institution scholar, examines Anthropic CEO Dario Amodei’s recent essay, which highlights a clash of interests. Amodei warns that AI advances are dangerous yet continues to pursue them, seeking protection from copycats while acknowledging that leading models were trained on others’ data without consent. He proposes shifting AI standards control from individual firms to a government‑backed club of companies and prioritizing a subset of nations—either the U.S. and its allies (17% of global population) or democracies (45%).

These ideas break traditional business rules that once emphasized total intellectual‑property control and worldwide reach. The article notes that the 1990s “end of history” ideal of the perfect company—maximizing geographic scope, minimizing diversification, owning core activities fully, and outsourcing non‑core functions—dominates Western markets today. However, rising nationalism, AI integration into physical goods, and supply‑chain disruptions are reshaping corporate strategy.

Governments now favor domestic firms, making domestic expansion more profitable, while asset managers are pressured to be “patriotic.” Historical examples from 1914 and 1936 show earlier periods also adapted business models to geopolitical turmoil. Companies like Sberbank in Russia, Huawei in China, and firms that retained exit options in Russia illustrate how control is being re‑defined. In the U.S., even giants like Amazon and Alphabet are evolving into conglomerates, reflecting a broader shift away from the old golden rules.