Fwiw, ElevenLabs doesn't even make any sense as a firm, considering its TTS offering is extremely overpriced by an absurdly large factor. A large user would be far better off using local models. Even the OpenAI TTS is cheaper. There is no moat here such as with hardware.

The moat is marketing and good looking UI. They have been spamming Tik Tok et al. relentlessly and every lazy opportunist algorithm exploiter out there knows who they are, what they do, and that they can easily generate enough slop per cost to make their money back. So yeah, not a moat, but you're kinda missing the point. Just like Lovable is wildly over-priced, under powered, and has crazy levels of lock in built in (you can't even access your own supabase account that it uses for backend). It's not for you / us. Its for the masses who don't know (or want to know) any better.