A relatively small Chinese artificial intelligence (AI) start-up created a large language model (LLM) generative-AI chatbot to rival OpenAI’s ChatGPT. The major story is the bomb the company dropped in January: that it reportedly trained its model using just $6 million and a couple thousand Nvidia (NASDAQ: NVDA) graphics processing units (GPUs). Meanwhile, U.S. tech companies spend billions and order hundreds of thousands of GPUs for similar work.