
Jul 31, 2026 · 46 min
AssemblyAI processes 120 million weekly voice conversations as infrastructure demand surges
AssemblyAI CEO Dylan Fox on 120 Million Voice Conversations a Week
As voice interfaces transition from simple transcription to context-aware reasoning, the underlying infrastructure must scale to support hundreds of millions of real-time conversations.
- 1AssemblyAI processes over 120 million voice conversations weekly by focusing strictly on the infrastructure layer instead of applications.
- 2Voice technology is expanding beyond simple transcription into context-aware models that power coding agents and robotics.
- 3The rapid rise of realistic AI voice agents requires new frameworks for sovereign deployments and ethical disclosure.
Don't miss
Dylan Fox details how AssemblyAI operates internally as an AI-native company, including his personal use of an AI agent.
The brief
AssemblyAI CEO Dylan Fox joins host Molly O'Shea to discuss the massive scale of his voice AI infrastructure platform, which now processes more than 120 million voice conversations every single week.
Since launching in Y Combinator's first AI batch in 2017, the company has focused strictly on the infrastructure layer rather than building applications, betting on the long-term demand for highly efficient, context-aware speech models.
While the future of human-computer interfaces may not completely eliminate the keyboard and mouse, voice is rapidly expanding as a reliable data-capture tool for everything from enterprise coding agents to advanced robotics.
As AI voice agents become highly realistic, the industry faces critical challenges around multilingual training data, sovereign self-hosted deployments, and the ethical necessity of clear AI voice disclosure.
Featuring
Books & mentions
Project Hail Mary
The sci-fi novel discussed by Molly and Dylan during their conversation on futuristic communication interfaces
Listen to the full episode and explore every guest, topic, and moment on PodLume.

AssemblyAI
Human-computer interface
Robotics
Sovereign AI