AFAIK, it is well-known in the industry that Tesla is the only relevant vendor (or at least one of the few) betting on video-only. A couple of years ago, this looked (while risky) more reasonable, given the hype around deep learning-based image recognition (combined with the assumption that humans primarily rely on video). But I suppose most vendors assumed even then that video-only won't be able to handle all edge cases, like bringing a car to a safe stop in case video is non-workable in an extreme situation.