VC10X - Why the Labs Suddenly Love Open Weights - David Epstein, GP, USF Ventures

This is more in the startup world, kind of they claim, "We have 100 employees, 10 of which are human."

That's interesting. I didn't know that's ac- already happening [chuckles] in the ecosystem.

[laughs]

People are reporting headcount numbers like that now. Amazing.

Yeah, headco- AI headcount as well. [laughs]

The open source AI letter is out, and a lot of the big companies, or most of the big companies have actually signed it, including OpenAI, except Anthropic. How do you look at that?

I think it is defensive. I mean, I think they, they realize exactly what I was saying before. They realize that companies are not going to be spending the money if they can get cheap models that do, um, as well for certain tasks. You know, it's the idea that, uh, the, the perfect is the enemy of the, of the good.

We continue to see layoffs from large and small companies, uh, with some blame on AI. Uh, what do you think about that?

I think this is just the beginning.

Hey everyone, this is Prashant Tandon, your host of the VC10X podcast, and today we have David Epstein with us. David is a general partner at USF Ventures and was previously a general partner at Crosslink Capital. He's done more than half a dozen startup. He teaches at the University of San Francisco, and he began his career designing computers at Data General. In this conversation, Dave makes the case that the real AI bottleneck isn't chips or power, it's data. He explains why frontier labs backing open models is a defensive move, not a principled one. And he tells me why he thinks jobs will disappear faster than they get created. So without wasting any time, let's dive straight in. This episode is brought to you by PodcastNX, my podcasting agency for VCs. We help VCs start and run branded podcasts that can serve as their entire content engine. Turn the conversations you're already having into a machine. One recording becomes your presence on LinkedIn, Spotify, YouTube, and email. Don't just build another VC firm. Build a firm that can stand out to LPs and one that founders can brag about having you on their cap table. Check out our services at podcastnx.com. I'll put the link in the description below. Now, enjoy the episode. Hey David, so good to have you on the VC10X podcast. How are you doing?

Thanks, Prashant. Yeah. Ver- I'm doing well, and, uh, I just wanna say that I've enjoyed your, uh, podcast in the past. They're insightful and current and interesting, so I'm excited to be on.

Absolutely. My pleasure having you on, and, uh, really thank- thanks for the compliment. It means a lot to me. Uh, so to start things off, let's talk about USF Ventures. USF Ventures sits at the heart of Silicon Valley's innovation ecosystem. From your vantage point, what feels fundamentally different about the AI startup landscape today compared to just two years ago?

Uh, interesting question. So, uh, AI, I don't think is a, uh, no longer a thing that's independent of companies. I think that, uh, we right now, maybe we can characterize frontier and foundation model development companies as being AI companies independent. But now it's not a novelty anymore. It's, it's very normalized. Every company has to be using and is