What "stronghold"? The science of LLMs is widely known, with everyone adding new small tricks everywhere. The main moat so far has been the availability of comp

What "stronghold"? The science of LLMs is widely known, with everyone adding new small tricks everywhere. The main moat so far has been the availability of compute, and that's a leaky moat because inevitably more companies will find enough compute or make snappier models, or simply take more time to train them. There is no secret sauce or moat , they ve been saying that for years, and it's still true. Being number 2 in the race is a losing position, so china probably thought they might as well give it away