Keep pulling the thread on Jordan Schneider.
The Chinese government has banned the deployment of Western closed-source AI models within China, creating a protected market for domestic companies.
The US policy to control exports of AI chips and semiconductor manufacturing equipment to China was implemented in October 2022, one month before the public release of ChatGPT.
The Trump administration approved the sale of NVIDIA's H20 chips to China, but the Chinese government subsequently rejected the offer.
During a summit in Seoul, President Trump was reportedly persuaded by advisors Howard Lutnick, Marco Rubio, and J.D. Vance not to approve the sale of advanced AI chips to China.
NVIDIA CEO Jensen Huang's primary argument for selling chips to China is that it keeps Chinese engineers developing within the CUDA software ecosystem, thereby preventing them from strengthening a competitor like Huawei.
China's domestic semiconductor industrial policy project is the largest in human history by investment amount and is not capital constrained.
There are currently no US government restrictions preventing Chinese companies from renting compute power, including top-tier NVIDIA chips, from cloud providers located outside of China.
The primary manufacturing constraint for China's semiconductor industry is the lack of access to EUV lithography machines, which effectively caps their domestic production at the 7-nanometer process node.
The Chinese government has subsidized its domestic hardware ecosystem with hundreds of billions of dollars over the past 15 years to build resilience against export controls from companies like TSMC, Tokyo Electron, and ASML.
Chinese technology firms can access top-tier NVIDIA chips and compute infrastructure located outside of China, with ByteDance reportedly using Oracle's US-based cloud for its operations.
Around 2021-2022, the Chinese government jailed the top seven leaders of its $100 billion semiconductor industrial policy fund for corruption.
The Western semiconductor supply chain, including companies like TSMC, has a manufacturing capacity lead of approximately 10x to 15x over Huawei's current domestic capabilities.