Bank of America analyst Matti Zhao stated that neither the US nor China can achieve AI dominance independently due to global supply chain interdependencies. While China excels in energy infrastructure and production equipment, it remains reliant on foreign advanced chips and memory. Conversely, the US faces energy grid limitations, with infrastructure averaging 30–40 years old compared to 15–20 years in China.
The US faces potential power capacity constraints for expanding AI infrastructure, while China requires breakthroughs in advanced chip manufacturing. Global partners like South Korea, a leader in memory production, and Japan, a supplier of high-tech materials and equipment, remain essential to the industry. China is also a major infrastructure supplier, with transformer exports reaching ¥64.6 billion in 2025, a 30% annual increase.
Despite ongoing trade tensions and efforts to reduce mutual dependence, the AI industry remains globally connected. Recent diplomatic developments include plans for a bilateral AI dialogue and a trade body between the US and China. However, technological rivalry persists, with Washington raising concerns over model distillation and Beijing criticizing US restrictions on advanced AI systems.
Data from the Rhodium Group indicates that the combined annual revenue of the seven largest Chinese AI models is approximately 10% of that generated by US-based OpenAI and Anthropic. This disparity highlights the current market landscape, where competition and cooperation continue to coexist as both nations navigate their reliance on international resources, including energy and cooling systems.
