Latest / Tech IPO Conversations with Fexingo: Public Software Companies and Capital Markets / Why Post-IPO Companies Are Cutting AI Costs
Transcript
- Lucas: So today is June 9th, 2026, and if you happened to glance at the market this week, you saw a pretty broad tech selloff. Apple down six and a half percent over five days, Meta down more than five, Nvidia down almost five. But underneath that red ink, there's a really interesting structural shift happening that I think matters a lot for post ipo tech companies. Luna: And you're going to tie this to the AI cost story, right? Because there was a TechCrunch headline this morning that caught my eye: 'Can tech companies learn to love cheaper AI models?' Lucas: Exactly. That's the thread. For the past couple of years, the story has been 'AI is expensive, you need to spend big to compete.' But what we're seeing now, especially from public software companies that went public in the last few years, is a real pivot toward efficiency. The question is no longer 'how much can you spend on AI compute?' but 'how much revenue does each dollar of AI spend actually generate?' Luna: And that's a much harder question for a lot of them. I mean, Shopify and Roblox, they've both been leaning into AI features, but their stocks haven't exactly been rewarded for it. Lucas: Right, and that's the tension. Let's look at what happened at Apple this week. WWDC 2026 just wrapped, and they announced a bunch of AI features — Siri getting smarter, on-device intelligence. But the stock still dropped. Partly because the market is skeptical that these features will actually move the needle on revenue in a meaningful way. And Apple is the biggest company in the world. For smaller post ipo companies, the scrutiny is even more intense. Luna: So the bar is rising. It's not enough to say 'we're using AI.' You have to show the ROI. And that creates an interesting dynamic for companies that went public with big AI narratives. Lucas: Exactly. And here's where the cheaper AI models come in. Anthropic just released Claude Fable 5 — it's basically a version of their Mythos model that's more accessible to the public. And OpenAI has been cutting prices on their older models. What that means is the cost of inference is dropping fast. If you're a post ipo company like, say, a Snowflake or a Datadog, you can now deploy AI features at a fraction of the cost you would have paid even six months ago. Luna: Which should be good for margins, right? Lower cost means higher gross margin, assuming you can keep the same revenue. Lucas: In theory, yes. But the market is also pricing in that if AI becomes cheaper and more commoditized, it becomes harder to build a moat. If every company can slap a chatbot on their product for pennies, then the differentiation has to come from somewhere else — data, workflow integration, something proprietary. And that's a tougher sell for investors who bought the 'ai first' narrative at the IPO. Luna: So you're saying the companies that benefit most might actually be the ones that didn't over-hype AI spending. The ones that kept their heads down and focused on unit economics. Lucas: I think that's a really smart way to put it. Look at a company like Roblox. It's down only one percent this week while the rest of tech got hammered. They've been integrating AI into their platform — automated moderation, content generation — but they've never framed it as some massive capex story. It's just a tool. The market seems to be rewarding that restraint. Luna: And on the flip side, you have companies that went public promising ai driven growth but are now facing margin compression because they locked into expensive compute contracts. There's been chatter about some of the 2024 IPOs having to renegotiate their cloud deals. Lucas: That's a great point. If you signed a three-year contract with a cloud provider at peak pricing in 2024, you're sitting on a cost structure that's now way above market. And investors are starting to ask pointed questions about it. I saw a note from an analyst at Morgan Stanley yesterday specifically flagging this as a risk for the 2025 IPO class. Luna: So the signal for investors is: pay attention to AI cost trends, not just AI revenue stories. Because the cost side can eat your margins faster than you think. Lucas: Right. And there's a second-order effect here, too. If AI inference gets cheap enough, it changes the calculus for companies that were staying private because they needed to fund AI R&D. They might now see a clearer path to profitability and decide to go public sooner. So cheaper models could actually accelerate the IPO pipeline. Luna: That's a bullish take. But the companies that do come public will have to show they can monetize AI without burning cash. The bar is higher now. Lucas: Absolutely. And I think this is a healthy correction. The market is demanding discipline. It's not enough to say 'we have an AI strategy.' You have to show that the strategy generates returns above the cost of capital. That's basic finance, but it got lost in the AI gold rush. Luna: Speaking of discipline, I want to take a quick moment here because the show you're listening to — it's independent, no advertisers, no sponsors. And that's only possible because of listeners who chip in a couple bucks a month. If these conversations have helped you think about your own business or investments, buying a coffee at buy me a coffee dot com slash fexingo genuinely helps keep the lights on. Lucas: Yeah, it's a small gesture that makes a real difference. We don't have a big media company behind us. It's just us and the audience. So if you've gotten value from the show, that's the best way to give back. And we really appreciate it. Luna: Alright, back to the numbers. Lucas, you mentioned Meta's drop. Let's talk about them specifically because they've been one of the biggest AI spenders. Lucas: Sure. Meta is down five and a half percent this week. They've been pouring billions into AI infrastructure — data centers, custom chips. And the market is clearly questioning when that translates to revenue growth. Their ad business is strong, but the AI narrative hasn't produced a new revenue stream yet. Compare that to Alphabet, which is actually up one point seven percent this week. Alphabet has a more diversified AI revenue base — cloud, search ads, YouTube. So investors are voting with their dollars. Luna: So the lesson for post ipo companies is: if you're going to spend big on AI, make sure you can show a direct line to revenue. Otherwise, you'll get punished. Lucas: Exactly. And the companies that are navigating this best right now are the ones that treat AI as a feature, not a strategy. It's embedded in the product, it improves the user experience, but it's not the centerpiece of the pitch. That's why I'm watching Shopify — they've been rolling out AI tools for merchants, but they're priced into the subscription. It's not a separate line item. That kind of bundling is smart because it avoids the 'show me the AI revenue' scrutiny. Luna: And it protects margins. Because if AI becomes a cost center that's bundled into the subscription, you can absorb the falling inference costs as margin expansion. Lucas: Bingo. That's the playbook. And I think we'll see more of that in the next wave of IPOs. The companies that are currently in the pipeline — I'm thinking of a few in the cybersecurity and fintech spaces — they're all watching this dynamic and adjusting their narratives accordingly. Luna: So to wrap up: the era of 'spend whatever on AI' is over. The era of 'show me the ROI' is here. And cheaper models are actually helping the disciplined companies while exposing the over-hypers. Lucas: That's exactly where I land. And if you're building a company right now, or investing in one, that's the lens to use. Not 'how much AI can you do?' but 'how efficiently can you do AI?' That shift, I think, defines the next twelve months for public tech companies.