Latest / Tech IPO Conversations with Fexingo: Public Software Companies and Capital Markets / Why Tech IPOs Are Hiding Their World Models
Transcript
- Lucas: There is a quiet but accelerating shift happening in how public software companies handle their most valuable assets. If you look at the broader market right now, specifically in the artificial intelligence space, the old playbook of hyper-transparency is crumbling. Luna: You mean the days of explaining every line of code to shareholders are over? Lucas: Exactly. For years, if you went public, you had to prove your moat was real. You showed the data, you showed the architecture. Now, companies are starting to treat their core AI engines—specifically things like world models—as trade secrets that belong in the vault, not on the slide deck. It’s a fundamental change in investor relations. Luna: That sounds like they’re protecting themselves from competitors more than they’re informing investors. Lucas: It’s both, but the investor angle is where it gets messy. Take Palantir, which is currently trading around one hundred seventy-seven dollars. They’ve been remarkably open about their platform. But even they, along with several other major players, are beginning to describe their underlying model capabilities in vague, strategic terms rather than technical specifics. The headline from TechCrunch today about world model companies keeping secrets isn't just gossip; it's a signal of where the industry is heading. Luna: So investors are essentially buying into a promise of capability without being able to verify the mechanics? Lucas: Precisely. And that creates a new type of risk. When you buy a software stock, you’re betting on future cash flows generated by those models. If the company refuses to disclose how the models are trained, or what data they’re using, you’re flying blind. You have to trust management’s word that the engine works, rather than seeing the blueprints. Luna: But isn’t that just standard corporate behavior? Why make a big deal out of it now? Lucas: Because AI is different. In the past, if a bank wanted to hide its lending algorithm, nobody really cared as long as the loans got paid back. But in AI, the algorithm the product. If you can’t see the product, you can’t assess its durability. We’re seeing a divergence where some companies are doubling down on openness to build trust, while others are retreating behind walls of proprietary secrecy. Luna: Which approach do you think wins in the current market environment? Lucas: I think the market is punishing opacity right now, but only for companies that don’t have a dominant position. If you’re the clear leader, like maybe NVIDIA with their hardware ecosystem, you don’t need to explain much because the results speak for themselves. But for pure-play software companies, hiding your AI strategy is becoming a liability. Investors want to know you’re winning, not just you’re winning. Luna: So it’s less about hiding and more about proving you’re not just another wrapper around someone else’s model? Lucas: Spot on. That’s the key distinction. Everyone is worried about being a wrapper. So companies are scared to show their work because they fear it will look thin. But by staying silent, they invite skepticism. It’s a lose-lose dynamic unless you have massive scale to fall back on. Luna: Does this mean earnings calls are going to get even more frustrating for analysts? Lucas: They already are. But this will make it worse. You’ll hear more language about 'strategic partnerships' and 'proprietary infrastructure' instead of actual metrics on model performance or data quality. It forces analysts to rely on third-party data or customer surveys to gauge health, which is always lagging and incomplete. Luna: And that lagging data could lead to sudden repricing events when the truth finally comes out. Lucas: Absolutely. Volatility increases when information is asymmetric. We might see sharper drops for companies that were hiding weaknesses, but also bigger rallies for those that suddenly decide to pull back the curtain and prove their edge. It’s a binary outcome for many mid-cap tech stocks. Luna: What about the regulatory side? Isn’t there pressure for more transparency in AI usage? Lucas: Regulators are asking for it, but they’re not demanding it yet. The SEC focuses on financial materiality. Unless a specific model failure causes a direct financial loss, it’s hard to argue that the architecture itself is a material fact required for disclosure. But that’s changing. As AI becomes central to operations, the line between operational secret and financial risk is blurring. Luna: So we’re moving toward a era where 'we can’t tell you' becomes a red flag rather than a badge of honor. Lucas: Yes. In the early days of the internet, secrecy was a virtue. In the age of AI, transparency of outcome is becoming the virtue, even if the input remains hidden. Investors need to see the results clearly enough to judge the sustainability. Luna: Are there any companies getting this right currently? Lucas: Some are trying. There are a few smaller platforms that publish detailed benchmarking reports on their model accuracy and latency. It’s risky—they’re giving competitors free data—but it builds immense credibility with sophisticated investors. It signals confidence that their tech is superior, not just different. Luna: That takes guts. Most CEOs would prefer to let the market guess. Lucas: It does take guts. And it requires a level of maturity in the investor base that isn’t always present. But as the sector matures, I expect more companies to follow that path. The noise is too high otherwise. Luna: Do you think this affects how these companies price their initial public offerings? Lucas: Huge effect. If an IPO candidate is secretive about its AI core, underwriters will have a harder time building a compelling narrative. They’ll either have to inflate the hype elsewhere or accept a lower valuation multiple. Secrecy compresses multiples because it adds uncertainty premium. Luna: So the most transparent startups might actually command higher valuations at the IPO gate? Lucas: Potentially, yes. Provided that transparency doesn’t reveal a fatal flaw. It’s a delicate balance. But generally, clarity reduces the discount investors apply for unknown risks. Luna: It’s fascinating how the dynamics of trust are shifting in real time. Lucas: It is. And it’s going to define the next cycle of tech investing. If you can’t see the engine, you better believe in the driver. Luna: Speaking of drivers, I appreciate how you guys keep this show running without ads. Lucas: Yeah, it’s something we talk about occasionally. A small group of listeners supports us monthly through buy me a coffee dot com slash fexingo, and that literally funds the production costs. Luna: It’s great to have that direct connection instead of relying on ad revenue. Lucas: It keeps us honest. We don’t have advertisers telling us what to say, so we can dig into topics like this without worrying about offending a sponsor. Luna: And it lets us cover the nuanced stuff that might get glossed over in a sponsored segment. Lucas: Exactly. So if you find value in these deep dives, checking out that link is the best way to support the network directly. Luna: Back to the topic though—do you think this secrecy trend will lead to more M&A activity? Lucas: I think it will. Public companies will acquire private AI firms precisely because they want access to the technology without having to compete against them in the open market. It’s a way to secure the secret before the public ever sees it. Luna: So the acquisition becomes the ultimate form of due diligence? Lucas: In a way, yes. You get to see the black box up close before you buy it. For public markets, it’s safer to own the asset than to bet on the competitor who might be developing it. Luna: That makes sense. It’s a defensive play wrapped in an offensive move. Lucas: Exactly. And it raises the stakes for everyone else. If you’re not part of the M&A wave, you’re stuck competing in a market where your rivals have already secured the secrets. Luna: How should individual investors adjust their strategy given this shift? Lucas: Focus on outcomes. Don’t get lost in the technical jargon of the pitch. Look at retention rates, look at gross margins, look at customer expansion. If the AI is working, the financials will show it regardless of how much they hide about the code. Luna: So ignore the hype and watch the cash flow? Lucas: Always. Cash flow doesn’t lie, even if the press release does. If the revenue is growing and churn is low, the model is probably solid. If you’re chasing the story, you’re vulnerable. Luna: Simple advice, but crucial in this opaque environment. Lucas: It is. And as the sector evolves, that discipline will separate the winners from the washouts. Luna: One last question—if secrecy becomes the norm, do you think we’ll see a backlash from activist investors? Lucas: Almost certainly. Activists thrive on exposing inefficiencies. If a company is hiding too much, activists will demand audits, demand disclosures, and demand changes. It could become a major battleground in shareholder meetings. Luna: So the fight for transparency might come from inside the house. Lucas: Yes. And that’s a healthy sign. It means the market is correcting for the lack of information. We’ll see more proxy contests focused on disclosure policies in the coming years. Luna: Interesting to think about the governance side of AI. Lucas: It’s the next frontier. How we govern what we can’t fully see will define the stability of these markets. Luna: Let’s see if the boardrooms catch up to that reality soon. Lucas: They will. The pressure is mounting, and eventually, the secrets won’t be worth the cost.