Could Walmart train large language models—and evaluate them well?

One post asks “Why not Walmart?” after pointing to Chinese companies’ ability to train large language models. It claims food-delivery company Meituan was the first to train a model with more than 1 trillion parameters on Ascend hardware, and says JD and China Telecom can train “serious LLMs.” A reply raises a separate question: would Walmart be a good model evaluator?