ElevenLabs — AI supply-chain exposure

ElevenLabs is a producer of multimodal AI, focused on the audio side — synthetic speech and voice agents. Its text-to-speech and voice APIs turn written text into natural-sounding speech and let other products give software a voice, which has made it one of the larger commercial audio-AI workloads and carried it to substantial recurring revenue. It sits in the applications layer, packaging generative models into a service developers call rather than run themselves.

The structural hook is that voice is a distinct, defensible slice of the model layer. General-purpose frontier labs chase text and images; ElevenLabs went deep on audio quality and a growing library of voices, building an API that others embed. That depth is its moat, but it also sits above the compute it consumes — every call it serves is demand flowing down to the accelerators and clouds beneath it, which is why the model reads it as an application producer rather than an infrastructure supplier.