Investing in ElevenLabs: Leading the Voice AI Modality

Ever since ChatGPT launched in November of 2022, we’ve seen knowledge work transform at breakneck speed. Nearly four years later, any credible business leader you can find is rushing to integrate AI into their operations, into their employees’ workflows, and entering a state of frenzied paranoia that they are not adapting fast enough.

Within the past week alone…

• Andy Jassy (Amazon CEO) has called generative AI the “biggest technology transformation in our lifetimes”
• Larry Fink (BlackRock CEO) has said that “demand for computing power is so large that a new asset class will spring up around it”
• Memory stocks like Micron and SanDisk have gone parabolic due to near-infinite demand from the AI data center use case and finite supply
• 75% of US GDP growth in the first quarter of 2026 was due to the AI boom, according to BEA (Bureau of Economic Analysis) data

The vast majority of token generation today is for text: think coding/software development, multi-step agents and automations, and the “talk to AI” chatbot/AI assistant use case. Most people, when they think about integrating AI into their operations, are thinking about text.

Get Mike Marg’s stories in your inbox

But what about other modalities of AI? Our economy doesn’t just run on text, it also runs heavily on voice. Voice is a massively important economic driver, and it’s both difficult and depressing to imagine a scenario where the business world ran 100% on text messages, typing, and code.

Prior to joining Craft, I worked in go-to-market roles at Dropbox, Slack, and Clearbit, and at each of those companies, success in my role was probably 80% due to the quality of my conversations with existing and prospective customers, and 20% due to the written work around those conversations (ie, email, CRM entry, etc.) Voice is the native modality of the business-to-customer interaction.

There is not a company on earth doing more for AI voice than ElevenLabs. First, they are creating some of the most reliable, beloved, and expressive text-to-speech (and speech-to-text) models on the planet. But beyond that, they are also productizing their models at a ridiculous pace to build creative tools, and business tools, for their users. There is a significant advantage in owning the underlying model AND productizing it for end users, and we’re seeing this play out right now in the insanely competitive LLM space.

ElevenLabs is also a classic bottom-up story (which we love at Craft) but for the AI age. A while ago, I wrote on my EarlyGTM blog that generative AI represented an evolution of the bottom-up SaaS model because of the crazy speed in which generative AI can start adding value to a user’s working life. With the previous generation of SaaS tools, you still had to spend legitimate time onboarding and figuring out the tool you were adopting. With generative AI, the best AI native products can create something extremely valuable for the end user with a simple prompt.

ElevenLabs has some of the best bottom-up, top-of-funnel traction (and self-serve traction) of any technology company we’ve ever seen. But beyond that, they are quickly building amazing products for prosumer creators, for professional creatives (who must deliver polished final products for marketing or entertainment use cases) and for business-to-business use cases (where companies want to use AI to speak with their customers.)

The b2b use case is an especially tricky job to be done and risks entering the uncanny valley- a customer has to leave any voice AI conversation feeling like it was just as good an experience as speaking with a real human. The bar for this type of interaction is very high, but that high quality bar represents an opportunity for a powerful moat. Even seemingly small differences in quality can make huge impacts in customer confidence and outcomes.

While ElevenLabs is probably most broadly known as a tool for creators (they just shipped some incredible music features this past month) they have a huge opportunity to build the modern customer support platform for the AI age, and are quickly executing on that opportunity. This is a massive market that pretty much any customer obsessed company cares about. But additionally, it is also a market that (we believe) will be won by the company that owns the underlying model and can deliver world-class products on top of those models.

We are thrilled to partner with Mati and Piotr, who just announced that ElevenLabs has crossed $500M of ARR after founding the company four years ago. While the world experienced the collective “ChatGPT moment” nearly four years ago, ElevenLabs is building the same groundswell for voice as we speak.

‍