DeepSeek Withholds Latest AI Model from US Chipmakers, Signaling Tech Decoupling

Chinese AI laboratory DeepSeek has reportedly decided to withhold its upcoming flagship model from major US chipmakers, including Nvidia, preventing standard performance optimization processes. The move marks a significant departure from established industry practices, where AI developers typically collaborate with hardware manufacturers to refine software-hardware synergy. This decision highlights the escalating technological rivalry and geopolitical tensions between Washington and Beijing. DeepSeek previously rattled global markets with its highly efficient, low-cost AI models, challenging the status quo of the industry. The lack of optimization for US hardware could potentially reduce the operational efficiency of Nvidia chips when running advanced Chinese AI architectures. Consequently, market uncertainty is rising regarding a potential decoupling in the global semiconductor and AI supply chains, impacting sentiment for stocks like NVDA and AMD.