AI Venture Risks: De-Risking & the $28 Billion Question #shorts

Name: AI Venture Risks: De-Risking & the $28 Billion Question #shorts
Uploaded: 2025-07-30T01:39:48+00:00
Duration: 1 min 37 s
Channel: Jason M. Lemkin
Description: In this short dialogue, Jason M. Lemkin and Rory discuss the core risk facing AI‑focused startups that rely heavily on large‑model providers such as…

30 Jul 2025 · 2 min

Watch on YouTube

1:37

In this short dialogue, Jason M. Lemkin and Rory discuss the core risk facing AI‑focused startups that rely heavily on large‑model providers such as Anthropic and OpenAI. Lemkin cites the Anthropic‑Winsor incident—where an M&A attempt was abruptly blocked—as a cautionary tale of platform dependence. They explore whether a venture‑backed company like Cursor should build its own model, secure secondary contracts, or negotiate binding agreements with multiple providers to mitigate that risk. The conversation then pivots to valuation: a $28 billion entry price implies investors are being paid for bearing the platform risk, and the team must decide if the massive market demand for AI‑powered coding tools justifies that gamble. Lemkin draws a parallel to the 2023 Shopify era, noting that attempts to go multi‑platform in B2B were largely futile, while acknowledging Klaviyo’s success as a counterexample. Finally, they outline practical de‑risking levers—licensing models, separate API contracts, building proprietary models, and multi‑contract strategies—suggesting that rapid revenue growth can afford startups the leverage to diversify away from a single AI provider.

Chapters

you should lean into this. The scary fact is what Anthropic did to Winsor, which is the minute you try and do an M&A, they cut you off at the knees and it hurt, right? So you probably, as Cursor, are saying, I can use all this momentum and all this venture dollars. Do I build my own model? Do I get a second source? Do I have to have a binding contract that applies to Anthropic? Do I have to have a second contract with OpenAI? You have to de-risk the big risk. But on the other hand, the prize is such that you don't just say, I can't build a business here because of this fundamental risk. You have to de-risk it because the core market demand you have is just so attractive. Do you think you're being paid for that risk if your entry price is $28 billion? That's a different question. You have to underwrite a $100 billion-plus company there. Well, you have to underwrite two things. I mean, listen.
It's a great— if we go back to 2023 before all of this, right? And imagine you're in a partner room and someone came to you with a deal like Herster and said, I've met these, these kids, they're so smart, right? But they're 100% platform dependent on another provider that will likely compete with them in the very near future and, and will raise infinite amounts of capital. Would you, would you have in 2023, would you have agreed to that deal? This is like Venture 101. You don't do these platform dependent. I did a bunch of stuff in Shopify. I know it's old school, but like what happened in Shopify is everyone tried to go multi-platform, Rory, and it was pointless because Shopify is 99% of the B2B market share.
I agree, but that's, for example, but there's a bunch of differences there. First of all, Klaviyo made it work. And secondly—
For now.
Yeah. And Shopify wasn't just a backend partner, it was a distribution point too. But I'll still take on your point and take it. Would you do that deal? And this is why the $28 billion is interesting. What you say to yourself is, if the core giant sucking sound of demand is so strong over the next 2 to 3 years, then the forward momentum, you're getting to a billion dollars faster than almost any other software company out there, is probably enough to allow you to have options. So, it's a calculated gamble, right? You roll the dice. And in this case, as you've so eloquently pointed out, the AI, the magic of AI in coding is so strong that you've got that kind of lift. Because even though you've still got this big existential risk out there, you've got more leverage now. There's a bunch of different things you can do. You're seeing licensing whereby you say, here's for the product, and then you get your API separately. You're definitely going to see people building their own models. You're definitely going to see multi-contracts to some extent. So, there's going to be a lot of de-risking happen. But you are at the founding stage, despite the platform risk, you're being paid for the risk.