Scott Stephenson

Scott Stephenson is co-founder and CEO of Deepgram, the speech company he started after leaving particle-physics research. He builds developer-facing speech systems that other teams meter into products, and he still talks about the work in the language of waveform analysis and first principles he learned underground in dark-matter labs.

Searching for Dark Matter

The lab was "basically in a James Bond lair," he told Madrona in a Featured Leader conversation: a liquid-xenon dark-matter experiment roughly two miles below the surface. He had come to it through academic physics, with a bachelor's degree from the University of Missouri-St. Louis and then a master's and PhD in particle physics at the University of Michigan, where he wrote analysis software that hunted rare event signatures inside huge waveform datasets.

Once the detector was running there were months of data-taking with little to do but wait. With Noah Shutty, a collaborator from the same physics world, he built wearable recorders that captured voice notes, conversations, and ambient audio around the clock. The devices produced "over 1,000 hours of audio," and when they tried to find key conversations, they found "countless hours of silence and background noise," he told AiThority. In a RedMonk Conversation he tied the physics habit to speech directly: mostly uninteresting signal, a few moments that matter, and models that can find "a needle in a haystack of recorded audio."

Founding Deepgram

"What we were working on in physics was discovering the fundamental laws of nature," he told Madrona. "But working on Deepgram would be discovering the fundamental laws of intelligence." In 2015 he left a postdoctoral post at UC Davis and, with Shutty and fellow Michigan physicist Adam Sypniewski, founded Deepgram. They saw that even advanced companies still used "antiquated technology" for open-ended audio, and "it became obvious that we were the ones to build it." Deepgram joined Y Combinator's Winter 2016 batch.

Eighteen Months of Smile and Dial

"We go to the market. We just get punched in the face over and over because everybody is asking us for speech-to-text," he said in an Unusual Ventures interview about the company's first product, an audio search tool built on embeddings that investors had liked. Deepgram turned to the simpler thing customers wanted. On Madrona's Founded & Funded podcast he said the company had to "earn our license to learn in this market" by winning one domain first.

The next hard stretch was sales. In 2018, as an eight-person company, Deepgram trained no new models for almost a year and a half, because its model was already strong and distribution was what held growth back. "All we did for that year and a half was smile and dial," he told Madrona. "If your day isn't 90 or 95 percent stuff you don't want to do as a technical founder in the early years, you may be driving yourself right off that cliff."

Fielding Unsolicited Offers at Deepgram

"We weren't out looking for a raise, though. We had multiple people coming to us," he told TechCrunch in January 2026, when Deepgram announced a $130 million Series C led by AVP at a $1.3 billion valuation. He said the company had been cash-flow positive the year before. With the round, Deepgram bought OfOne, a Y Combinator startup that built voice ordering for quick-service restaurants.

He has kept Deepgram focused on speech models that product teams buy as infrastructure: transcription first, then speech generation and tools for voice agents. Talking to Voicebot in 2021, he compared the approach to developer services: "Think of Twilio, think of Stripe, think of an API integration you use, which is a piece of a greater system you build yourself internally." On RedMonk he argued that voice is having its moment because speech-to-text, language models, and text-to-speech can now be joined into agents people will talk to.