DeepSeek has delayed the launch of its DeepSeek-R2 AI model, originally planned for May, after CEO Liang Wenfeng expressed dissatisfaction with its performance. The model aims to improve upon R1 with better coding abilities and multi-language reasoning support. Technical challenges are compounded by limited access to advanced Nvidia chips due to U.S. export restrictions, particularly the ban on H20 AI processors for China. While DeepSeek has shared technical specifications with Chinese cloud providers to prepare infrastructure, no new timeline has been announced. This delay reflects broader challenges facing Chinese AI developers amid global supply chain restrictions and technical hurdles.