DeepSeek founder: China lags US AI by 6–18 months, compute the gap

DeepSeek founder Liang Wenfeng attributed China's lag in artificial intelligence development to computing power constraints rather than a shortage of engineering talent. Speaking publicly, Liang said China's AI industry trails the United States by six to 18 months—a significant but not insurmountable gap. He emphasized that the shortfall stems primarily from limited computing power, not from any deficit in human expertise or research capability. This distinction is crucial: it suggests that China's path to closing the gap depends less on training more skilled researchers and more on acquiring or developing the hardware infrastructure—GPUs, TPUs, and data centers—necessary to train and operate large-scale AI models competitively. The range Liang cited, stretching from six months to 18 months, implies that the distance between the two countries is not fixed, but rather subject to fluctuations in compute availability, investment rates, and policy decisions around chip manufacturing and procurement.

Liang Wenfeng's statement reframes the US–China AI competition as fundamentally a hardware problem rather than a skills problem. By pinpointing compute as the decisive gap—not research acumen or engineering talent—he implicitly acknowledges that China has the human capital to advance AI systems, but faces structural constraints in accessing the vast computing resources required to train and deploy large-scale models. The six to 18 month range he provided suggests volatility in the competitive distance, contingent on how quickly compute capacity can be expanded or acquired. This framing has implications for supply-chain policy and semiconductor availability, areas where international geopolitics and trade restrictions directly shape competitive outcomes.

No comments yet. Be the first to share your thoughts!