Sound Ventures, the eight-year-old, Beverly Hills, California-based venture firm led by general partners Ashton Kutcher, Guy Oseary and Effie Epstein, announced a new $240 million AI fund just last week.

Turns out half of it is already invested in the three companies that it announced at the time that it rolled out the fund: OpenAI, Anthropic and Stability.AI.

Further, says Epstein, the team plans to invest in just six companies altogether, they aren’t setting aside money for follow-on rounds and backing rival companies is not, in the case of this fund, a deterrent.

The strategy flies in the face of traditional venture investing tenets, but the highly concentrated nature of its AI fund is a reflection of just how few power players Sound Ventures anticipates will emerge in a world remade by AI, including because of the technical talent required and the capital needed to cover compute costs. (OpenAI CEO Sam Altman has previously described these as “eye watering.”)

We talked with Epstein earlier today about the new, high-risk, high-reward strategy of Sound, which is separately investing an early-stage fund that it closed in 2021. Our chat, below, has been edited lightly for length.

TechCrunch: You said last week that you think just a handful of players are going to emerge as the victors here, given the talent and capital required to support these foundational model layers. You’ve backed three already — OpenAI, Anthropic and Stability.AI. Will you sink more into these same companies as they raise subsequent rounds? How highly concentrated will your AI fund be?

Epstein: So the fund will be quite a concentrated portfolio, and it’ll be concentrated across six to seven positions.

That’s wild. So, like, half the fund is gone already?

That’s probably a safe estimate. We’ve invested in three companies already — we’re not going to disclose how much we invested per company — but if you do the math, it’s about six positions, and three of them have already been made, so that’s pretty accurate.

Which company has attracted the most capital? Is it OpenAI?

We’re not going to disclose the amounts across each company but I will tell you the fund came together quickly from a fundraising perspective [thanks to the] work that we’ve done over the last 15 plus years [including when Kutcher and Oseary were investing their personal funds as angel investors]. And when you think of OpenAI specifically, Ashton and Sam [Altman] have known each other for over a decade. So we have fostered these incredible relationships with exceptional founders . . . and built a track record. And so that’s why we were able to get quite significant allocations when you look at these companies, in quite competitive rounds.

The team at Anthropic used to work at OpenAI and is ultimately trying to compete with OpenAI. Was there any pushback about your approach?

What’s so interesting about this ecosystem is that founders recognize they are all on a similar mission. Of course, it’s a competitive market. Companies are fighting for talent, as you’d want them to be, but at the end of the day, if we at Sound can help support them from a narrative perspective, from a branding and marketing perspective [and to share] what’s happening broadly within AI, that is something that benefits the ecosystem as a whole, and that’s why the founders were really supportive. It’s something that we would only do with the blessing of our founders.

What kind of information rights do you get from these companies?

I think the most important thing again is that we do this in collaboration with our founders. We’ve invested in more than 200 companies [since Kutcher and Oseary began investing in startups, including Uber, Airbnb and Spotify]. There have been cases in the past where we’ve invested in competitors as well, and that has been with the opt-in from founders on both sides. And our track record at the end of the day of supporting founders is the reason why we’re able to attract exceptional founders to Sound.

Has Sound talked to Elon Musk about what he wants to build? Would you potentially back that company if it comes to pass?

That honestly is not something I can comment on at this point. He is not part of the portfolio.

Okay. But you’re obviously aware that he’s trying to build a rival to OpenAI. 

In general, we’re aware of many of the players out there. That’s our job as venture investors. We need to understand what the market landscape looks like. [In fact] we’ve been investing in AI for a decade. We actually made our very first bet within artificial general intelligence back in 2007; we invested in Vicarious [the robotics and AI company acquired last year by Alphabet] and so we’ve been in and around this industry for a while and gotten to know some of the key players. It’s why this strategy came together for us quite quickly.

I appreciate this bet on the network effects at play here. Relatedly, there was a lot of attention paid last week to [the educational software company] Chegg, whose shares tanked after it acknowledged the degree to which OpenAI’s chatbot, ChatGPT, is disrupting its business. Chegg is a portfolio company of Sound as well. How do your companies protect themselves?

We’ve reached out to a number of our companies to help them understand and think through how they can leverage this type of technology internally. We believe that every company, regardless of industry or sector or the problem that they’re trying to solve, will have to figure out a way to incorporate this into part of their offering. I think something incredible that [Chegg CEO] Dan [Rosensweig] did was, in the few weeks before earnings came out, his announcement around a partnership with OpenAI.

How are you using AI inside Sound Ventures?

We use it for sourcing primarily, just to understand where talent is going. We’re so obsessed with talent.

In the meantime, you’re concurrently deploying another fund?

Yes. We have two strategies that we deploy in parallel. There’s the growth strategy, which is really dedicated to the foundational model here. And in parallel, we’re investing our [third, $200 million] early-stage fund. That is typically used to fund Series A and B software companies. As it relates to AI, that would mean funding the application layer that sits on top of the models. That’s an area we’ve been super, super active in and exploring deeply.