Trump Administration Authorizes Conditional Nvidia H200 AI Chip Exports to China

Great-power techno-export controls swing like a pendulum: in 1981 Washington let Toshiba sell sub-critical machine-tools to the USSR under safeguards—only to re-tighten controls five years later when the Soviet Navy used them to quiet submarines. Today’s H200 waiver resembles that moment: a carve-out meant to earn revenue and manage allies’ (or firms’) frustrations while claiming national-security vigilance. The announcement exposes two long arcs. First, the century-old pattern of the United States monetising, but ultimately trying to monopolise, breakthrough general-purpose technologies (from radio tubes in the 1920s to GPS chips in the 1990s) persists. Second, Beijing’s relentless push for self-reliance—mirroring Japan’s catch-up semicon strategy of the 1960s—continues; limiting purchases to “special circumstances” signals China views imports as a stop-gap, not a solution. Whether this moment matters in 2126 depends on pace: if domestic Chinese designs match or leapfrog U.S. GPUs within a decade, the 2026 waiver will be remembered as a brief revenue-grab that accelerated the rival’s climb; if not, it will look like a prudent, enforceable throttle that prolonged U.S. technological primacy. The articles, mostly Western wire copy, underplay enforcement difficulty and China’s capacity to repurpose lesser silicon, so the real strategic delta may only be clear years from now.