Latest / The Startup Exit Podcast with Fexingo: IPOs, Acquisitions, and Founder Liquidity Events / How AI Coding Startup Cognition Hit 25 Billion Pre-Money
Transcript
- Lucas: So Cognition Labs just raised one billion dollars at a twenty-five billion pre-money valuation — and I think that number deserves a slow blink. Twenty-five billion dollars pre-money for a company that makes an AI coding assistant. Luna: That is a staggering number. And they're not alone — earlier this month we saw OpenAI raise at what, three hundred billion? But Cognition is still relatively young. Lucas: Right, founded in 2023. They make a product called Devin — an autonomous AI software engineer. And the round reportedly came together fast. The lead investor is a crossover fund that typically buys public equities, so you have to wonder: does this signal a near-term IPO, or is this just the market pricing AI scarcity? Luna: Let's talk about that pre-money. Twenty-five billion — that's more than the market cap of some public software companies. Shopify, for comparison, is trading around one hundred seven dollars a share, market cap around one hundred thirty billion. So Cognition is valued at roughly twenty percent of Shopify, with a fraction of the revenue. Lucas: And Shopify has been around since 2004. Cognition is three years old. The revenue figures are private, but most estimates put Devin's annual recurring revenue somewhere in the low hundreds of millions. So you're paying a massive premium for growth potential and the thesis that AI will displace a huge chunk of software engineering. Luna: That thesis is either brilliant or dangerous. I mean, if Devin really does replace a mid-level engineer, the total addressable market is enormous. But we've seen this movie before with other AI startups that raised at astronomical valuations and then struggled to justify them. Lucas: Exactly. And this is where the exit angle gets interesting. At twenty-five billion pre-money, the IPO bar is incredibly high. You'd need to show a clear path to a fifty or sixty billion public market cap to make the late-stage investors whole. That's a very small club. Luna: So what are the options? IPO, acquisition, or stay private longer and do more secondaries? Lucas: Let's take acquisition first. Who buys a twenty-five billion dollar private AI company? The usual suspects — Microsoft, Google, Amazon. But Microsoft just reported earnings, and they're spending heavily on AI infrastructure. Google's parent Alphabet has been investing in its own models. And there's regulatory risk. The FTC has been skeptical of big tech buying AI startups. Luna: Right, we saw that with the Microsoft-Inflection situation. They structured it as a hiring deal, but regulators are watching. So an IPO might be the cleaner path, but the public markets have been choppy for high-growth tech. Look at Coinbase — trading around one seventy-six, down over eight percent in the last five days. Not exactly a welcoming environment. Lucas: Yeah, but AI names have been bid up aggressively. NVIDIA is still above two hundred dollars, though it's down almost five percent in the last five days. And Meta is up four percent in the same period. So there's appetite, but it's selective. Cognition would need to convince public investors that it's not just a feature that Microsoft or Google could replicate. Luna: And that's the existential question: is Devin a standalone platform or a feature of a larger IDE? We've seen GitHub Copilot, which is essentially Microsoft's answer. And Amazon has CodeWhisperer. The incumbents have distribution. Lucas: Which brings us back to the valuation. The twenty-five billion pre-money implies a certainty that Cognition will win. But in venture, a few big wins can cover many misses. The question for founders and early investors is: when do you take liquidity? Some early employees probably have paper worth millions, but until there's an exit event, it's just paper. Luna: And that's where secondaries come in. We talked a few episodes ago about how private secondaries have become a liquidity lifeline. Cognition could facilitate a tender offer, let employees sell some shares. That might be the smartest move — let people de-risk while the valuation is high. Lucas: You know, it's conversations like this that remind me why we keep this show ad-free. We don't have to pitch you between the analysis. If you find this kind of breakdown useful for what you're building or running, the best way to support that is a small contribution at buy me a coffee dot com slash fexingo. It keeps us independent and focused on the numbers that matter. Luna: Yeah, we really appreciate it. Just a simple way to say this content matters to you. Lucas: So back to Cognition. Let's look at the team. The founders are former AI researchers from Google and Amazon. They have a strong technical pedigree. But the CEO, Scott Wu, has said in interviews that they're building an 'AI teammate,' not just a tool. That framing matters because it suggests a higher price point and a different go to market. Luna: An AI teammate that costs, what, five hundred dollars a month per user? That's the rumored enterprise tier. If they can land large enterprise contracts, the unit economics could be very attractive. But enterprise sales cycles are long, and companies are cautious about letting AI touch production code. Lucas: Exactly. Trust is the biggest barrier. One mistake in production could cost millions. So Cognition needs to build a track record. That's why the valuation feels like it's pricing in a future that hasn't happened yet. Luna: But let's play devil's advocate. If AI coding assistants improve at the current rate, within two years they might be able to handle a majority of standard programming tasks. That would be a tectonic shift in the software industry. And being the first to scale could create a winner take most dynamic. Lucas: That's the bull case. And the reason crossover investors are piling in. They see the trajectory. But as a startup exit podcast, we have to ask: how does this company become a liquidity event for its early backers? An IPO at a fifty billion valuation requires a certain revenue multiple. Even at a generous ten times forward revenue, that would imply five billion in ARR. They're not there yet. Luna: So maybe the exit is not an IPO but an acquisition by a hyperscaler. Let's say Microsoft buys them for forty billion. That would be a win for the VCs, but what about the anti-trust risk? The regulatory environment has changed. Lucas: It has. But these things are often resolved with concessions. And if the alternative is a Chinese AI company becoming dominant, regulators might blink. It's a geopolitical calculus. Luna: Interesting. So the valuation makes sense only in a world where AI transforms software development and the company either IPOs or gets acquired by a tech giant. Both outcomes are possible, but not guaranteed. Lucas: Right. And that's what makes this such a fascinating case study for startup exits. The mega-round sets a high bar. The founders and early employees are sitting on enormous paper wealth, but turning that into cash requires a market event. The smart ones will start thinking about secondaries now. Luna: We've seen that with other companies — Stripe, SpaceX. They used tender offers to let employees cash out without an IPO. Cognition could do the same. It keeps the company private longer, which might be beneficial if they need more time to hit those revenue targets. Lucas: Exactly. And with this round, they have a fresh valuation to set the price. So I wouldn't be surprised if they announce a tender offer within the next six months. It's a classic move: raise at a high valuation, then let insiders sell a small percentage to create liquidity. Luna: And for the late-stage investors, they get to see how the secondary market prices the shares. If there's strong demand, it validates the valuation. If not, it's a warning sign. Lucas: So to wrap up: Cognition's one billion raise at twenty-five billion pre-money is a bet on the future of software engineering. It's a high-risk, high-reward scenario that will be a test case for AI startup exits. Whether through IPO, acquisition, or private secondaries, the next two years will tell us a lot. Luna: And we'll be watching. Thanks for listening.