US‑China AI Compute Competition Shifts to Domestic Chips and Distributed Governance

Analysts note that the narrative of an AI arms race between the United States and China is increasingly missing a broader, more collaborative development. Research shows that dozens of international bodies—including the UN, OECD, WTO, NATO, APEC, ASEAN, the African Union and the G20—are independently converging on common AI‑governance goals such as algorithmic accountability, workforce adaptation, data protection, security cooperation and applications for climate and health. This distributed governance model challenges the view that AI policy is a zero‑sum game between two superpowers.

At the same time, the economics of AI compute are changing. U.S. export controls are curbing Nvidia’s advanced AI‑chip sales in China, driving Chinese demand toward domestic suppliers, especially Huawei, whose market share in China is projected to rise to roughly 50% while Nvidia’s falls to about 8%. OpenAI’s recent announcements—including a preview of GPT‑5.6, new agent capabilities and a custom inference chip built with Broadcom—demonstrate a trend toward vertical integration of the frontier AI stack. Additionally, U.S. grid regulators are treating large AI data centres as a distinct load category, requiring new interconnection rules and rate structures.