I've Seen This Movie Before: The Twin Illusions That Will End the AI Gold Rush
After three decades analyzing market cycles, today's AI boom feels like déjà vu. It's a gold rush fueled by two illusions—platform and data dependency. This is the pattern I've seen play out right before every bust.
After three decades of watching financial markets, you learn that every boom-and-bust cycle feels unique on the way up, but shockingly similar on the way down. I remember the dizzying heights of the dot-com bubble, where companies with no revenue were valued in the billions. I had a front-row seat for the subprime mortgage crisis, built on financial instruments so complex, few understood their own risk.
Today, I get that same feeling of déjà vu. The AI revolution is, without a doubt, a profound technological shift. However, the speculative frenzy it has unleashed in the world of SaaS and AI startups echoes the same irrational exuberance I’ve seen before.
Investors are pouring capital into a sea of startups, dazzled by slick demos and promises of exponential growth. But they're turning a blind eye to two fundamental risks that are hiding in plain sight. I call them the Twin Illusions: the illusion of Platform Independence and the illusion of a Data Advantage. And when the current gold rush ends, these two factors will be what separates the enduring companies from the digital ghosts.
Illusion #1: Building a Castle on Borrowed Land (Platform Dependency)
One of the most common pitches I see today goes something like this: "We've built a game-changing AI that integrates seamlessly with Microsoft Excel to automate financial modeling!"
On the surface, it sounds brilliant. It’s a massive market with a clear pain point. But what happens when they strike gold?
Let's play this out. The startup launches, finds product-market fit, and gets great press. Microsoft, the platform's owner, watches this unfold from its global headquarters. They see a validated market and a proven use case, all developed and tested with someone else’s venture capital. Then, in the next major Microsoft 365 update, they roll out a native, slightly better version of that exact feature.
Overnight, the startup's primary value proposition vanishes. They were never a standalone business; they were an unpaid R&D department for Microsoft.
This isn’t a hypothetical; it’s a pattern. Building your core business on another company's platform is like building a castle on borrowed land. You don’t control the access roads, you don’t own the mineral rights, and the landlord can change the lease—or decide to build their own castle—at any moment. Whether it's Google Workspace, Salesforce AppExchange, or the Apple App Store, the platform always wins.
Illusion #2: Fighting a Tank with a Toothpick (The Data Moat)
The second illusion is even more critical in the age of AI. The mantra we hear from every founder is, "Data is the new oil." And it's true. However, the crucial question investors often overlook is: Whose oil is it?
Most AI startups are working with the same publicly available datasets or the relatively small, siloed data they can collect from their initial customers. They claim their proprietary algorithm is the key. But an algorithm, no matter how clever, is not a defensible moat.
The real oil—the vast, sprawling reserves of proprietary data—has been collected for decades by the incumbents. Google has a two-decade head start on understanding human intent. Meta has mapped our social connections and interests. Amazon knows more about consumer behaviour than any company in history.
A startup can’t compete with that. Their model might be clever, but a model trained on trillions of data points will almost always outperform one trained on millions. The incumbents are coming to this fight with a fleet of tanks, and most startups are showing up with a toothpick, hoping their algorithm is sharp enough to make a difference. It’s not.
Even worse, these two illusions feed each other. The platform owners don't just own the land; they get all the data generated by the people living on it, creating a powerful, self-reinforcing cycle of dominance.
The Inevitable Flight to Quality
So, is this another dot-com bust waiting to happen? Not exactly. The internet was a real, foundational technology, just as AI is today. But the bust wiped out the companies that were all hype and no substance.
A similar "flight to quality" is coming for AI. The market will soon stop rewarding hype and start asking the hard questions. When it does, the capital will consolidate around two types of players:
- The Incumbent Giants, who own the platforms and the data.
- The Rare Startups with a truly unique and defensible data moat that can’t be easily replicated or absorbed.
For everyone else, the movie will end as it always does. As an investor, a job seeker, or a founder, now is the time for critical thinking. Before you get swept up in the next big pitch, ask those two simple questions: Is this a real business, or just a feature for a bigger platform? And do they have their own oil, or are they just drilling in someone else's backyard?
History doesn't repeat itself, but it often rhymes. And right now, the market is humming a very familiar, and very precarious, tune.
This article was the warning. The report below is the proof.
It’s a 23-minute deep dive into the case studies and market data that back up every claim I've made. It’s not for the casual scroller. It’s for the serious investor, founder, and strategist who prefers rigorous analysis over convenient hype.
The question is, are you one of them?
Vlaad Gajic