Washington’s New AI Chip Export Guidance Explained

🔑 Enhanced Key Takeaways

• •The new guidance issued on May 31, 2026, explicitly clarifies that export licenses are required for advanced AI computing items destined for entities whose ultimate parent company is headquartered in China or Macau, even if the receiving subsidiary is located outside these jurisdictions, thereby closing a perceived loophole that may have allowed substantial volumes of advanced chips to reach China-linked projects through overseas subsidiaries.
• •This May 31, 2026 guidance reaffirms a licensing requirement initially introduced on November 17, 2023, which some industry players may have believed was suspended after BIS announced in May 2025 that it would not enforce certain aspects of the January 2025 "AI Diffusion Rule."
• •The guidance specifically targets advanced processors such as NVIDIA's Blackwell and Rubin chips and AMD's MI350X class of accelerators, which will generally necessitate licenses when their ultimate destination is a China-headquartered firm.
• •Despite the recent tightening, a separate rule effective January 15, 2026, shifted the review policy for certain advanced computing semiconductors, including NVIDIA H200 and AMD MI325X, from a "presumption of denial" to a "case-by-case review" for exports to China and Macau, contingent on strict conditions like performance thresholds, a 50% volume cap relative to US sales, and independent US testing.
• •China's Ministry of Commerce has strongly criticized the US actions, accusing it of abusing export controls under the guise of national security, which they claim severely harms the legitimate rights of Chinese companies, disrupts international trade order, and destabilizes the global semiconductor supply chain.

🛠️ Technical Deep Dive

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline