Elon Musk said existing semiconductor fabs may not have enough capacity to support the expansion of AI across data centers, edge computing, humanoid robots and vehicles. He cited potential supply disruptions in Taiwan alongside a broader manufacturing ceiling as reasons Tesla Inc. and Space Exploration Technologies Corp. are pursuing Terafab, a planned semiconductor project that SpaceX filings say is being developed with Tesla and that Intel Corp. joined in April. Nvidia Corp., which relies on foundry partners for leading-edge processors, could face constraints even if demand for its AI chips continues to grow. Taiwan Semiconductor Manufacturing Co. Ltd. supplies chips to major AI companies, leaving the broader AI supply chain exposed to both geopolitical and production risks. Musk said Terafab would need to address not only GPUs but also memory and advanced packaging. The project’s long-term target is one terawatt of compute hardware annually, although timelines, milestones and capital spending remain undetermined. Nvidia CEO Jensen Huang separately said Musk may be uniquely positioned to execute the initiative, while stopping short of saying Nvidia would manufacture chips there.