20 Comments
User's avatar
Max Chamuel's avatar

Kinda disagree with the Kalshi example. Tarek got the idea interning at Goldman in 2016, when clients wanted to bet on Brexit and the election and the desk had to cobble together options bundles that only sort of tracked it. Luana did Bridgewater, Citadel and Five Rings. They were pretty much finance people.

The difference is they weren't the banks. Goldman was making money on those clunky hedges, so why would they fix it? Feels like that's the actual pattern, not that domain experts can't disrupt their own industry.

Shafayat Hossain's avatar

"Maybe coding was never the main bottleneck for startups in the first place?" That's the line I'd keep. Andrew Ng made a similar point recently: the bottleneck now is deciding what to build. Building solo with Claude Code, I get the code fast. The customer conversations don't speed up at all, and they're what decides whether the v1 is right.

Steven Pang's avatar

Does “skeptical on international founders” mean anyone who is not from the US?

Jordan Bishop's avatar

I’m curious to hear what a16z thinks about the advantages of being a solo founder who can ship a product that scales from an investment perspective. What is the value prop of venture capital to this archetype when the software product is increasingly bootstrappable?

Not every team or solo founder will be able to do this of course, but I’m curious what the VC world thinks about founders in this category.

Pierrick Bouffaron's avatar

Your point about domain experts got me thinking about when founders need that expertise. In a factory, an early prototype might reveal that the delay is in batch sign-off. If the problem is getting records out of older systems, you need someone who’s deployed industrial software. If the plant needs to change who can approve a batch, you might need someone who’s led that change before. A cheaper first build gives the founder a chance to find out which person to bring in. We synthetized our thoughts here: https://www.entropiasignals.com/p/ai-and-the-monday-test

Venture Capital Jobs's avatar

Would love to see fund who are solely focused on funding content creators working on new ideas

NIA's avatar

Notice that every one of these archetypes reduces to the same claim: the demo got cheap. Fine, true. But Chen half-buries his own best objection in the "why not" — coding was never the bottleneck. It was the million customer meetings, the hiring, the reliability grind after the demo. AI didn't touch those, and if anything it made a working v1 worth less as a signal. Also, "non-technical founder who can't code but is passionate about business" is just the a16z house archetype with a new hat on.

Joao Bosco's avatar

Why skeptical on international founders? There are so many successful international companies founded by people born outside of the US, including Kalshi, which was mentioned in the article.

Crosscurrents's avatar

Hoteliers not creating Airbnb makes a sharp counterpoint on domain experts, though AI might shift the odds a bit, since the insider who knows the workflow can now prototype a fix over a weekend. Defense and manufacturing could be where that matters most. Former soldiers and plant engineers see problems VCs rarely hear about, and the country could use more of them building companies. The solo founder case seems strongest for software with a short sales cycle. Anything that has to clear procurement, compliance or a factory floor probably still needs a second person doing the million meetings. Curious whether Speedrun has seen more applicants from those industrial backgrounds this year.

thegolftechguy's avatar

In the age of agreeable AI we might see more solo founders but scaling their companies will require them to consistently make good decisions. The value of cofounders is not just that they can bring expertise you don't have (which AI could replace) but also that they can challenge your thinking. Smart non engineers may have valid opinions on the tech stack and good engineers will have valid opinions on GTM etc. The risk with solo founders with AI is that they go down rabbit holes that they can't dig themselves out of.

Unpursued AI's avatar

On the solo founder "why not": in Ground's October 2026 study of PM and founder threads, in-person recruiting worked in 97 of 98 comments that discussed it, while cold email and calls worked in 73 of 137. AI makes the v1 cheaper, but finding the right people still seems to need a person showing up. Does Speedrun look at how a solo founder plans to cover customer time, or mostly at what they've shipped?

Olive Margin's avatar

Investors agreeing on what a founder looks like, and each round making the next easier, is the point most worth testing. At seed, where several small cheques have to line up, a familiar profile can stand in for evidence for longer than it should. Looking back at which follow-on rounds were raised on milestones and which on the profile would show how much the profile adds.

Nan Deng's avatar

The gap I would not collapse is the prototype and the thing that can take a write. The essay's own "why not" on non-technical founders is that a demo is easier now and the reliable product is not, especially once the category is a dev tool rather than a landing page. I would test that as a write path: the same prompt that scaffolds a CRUD app does not decide which request is idempotent, which row version wins, or what happens when the agent retries a payment that already succeeded. If coding was never the bottleneck, the new bottleneck is the check that a generated change is allowed to land, and a solo founder who can ship v1 still needs that gate before the first customer data is in the system.

Sam Weinstein's avatar

This is what I'm seeing/