Cambricon, a Chinese AI chipmaker, has expanded from inference into training, achieved commercial profitability, and increased 1,763.6% from its IPO price, according to a December 23, 2025 Substack article. The article argued that Elon Musk needed control over semiconductor manufacturing, AI accelerator architecture, and training capacity, with GlobalFoundries (GFS) addressing manufacturing, Cambricon addressing inference, and Cerebras (Private) addressing large-scale training. Musk did not secure Cambricon or an equivalent independent accelerator platform, leaving Tesla (TSX) and xAI dependent on Nvidia (NVDA), TSMC (TSM), and Samsung Electronics (005930.KS) for critical computing capacity. A direct Cambricon acquisition would now be politically improbable and financially expensive, with the strategic mistake being the failure to secure comparable capabilities earlier, before AI accelerator valuations and supply constraints accelerated. The article states that China is creating domestic alternatives to Nvidia, while U.S. hyperscalers are developing proprietary accelerators. Nvidia remains the market leader, but an increasing share of training and inference demand will be captured by regional platforms and custom silicon.