China's homegrown AI chips no longer lack a software ecosystem — they lack the factory capacity to make enough of them. That's the argument attributed to DeepSeek founder Liang Wenfeng, whose company builds some of China's most closely watched AI models.
According to finance.biggo.com, Liang argued that China's AI chips "don't lack an ecosystem, only production capacity," and that Nvidia's CUDA software "moat" — long seen as the American chipmaker's biggest advantage — is crumbling. Reporting summarized by MSN says Liang told investors that China's shift away from Nvidia hardware has reached a turning point, and that the country could break its Nvidia dependence within a year, with chip supply the main obstacle. Chosunbiz likewise reported that Liang is touting a "chip opening" as Nvidia's grip loosens while DeepSeek pursues artificial general intelligence.
The push has government muscle behind it. The Wall Street Journal reports that China is racing to rival America's AI chips, with a confidant of Xi Jinping leading the effort. According to the Journal, a Vice Premier warned Chinese AI companies that anyone who resisted using local chips was a "traitor" — a striking signal of how hard Beijing is pressing firms to buy domestic.
The broader supply chain is shifting too. Times Now reports that China's memory-chip makers have gained significant negotiating power as the domestic AI industry grows, even under US export pressure.
Why it matters: if the bottleneck really is manufacturing rather than software, China's path to competitive AI hardware becomes a question of building enough fabs — a challenge of scale and money more than one of engineering — which could reshape both the global chip market and the US-China technology race.