I imagine it's because there is a *massive* scramble to incorporate language models into existing applications and most of the time you need a vector database to do that. As a proxy for understanding the scale of this scramble, ChatGPT is the fastest growing software product ever by a pretty substantial margin.