Prominent voices in the AI software world, the people actually building and selling this stuff like Palantir CEO Alex Karp, are starting to sound a new alarm.
Their worry isn’t just “the bubble” — it’s a fundamental, dollars-and-cents business problem.
The warning? For huge parts of the AI market, the cost to build and implement the technology simply may not create enough value to justify the expense.
The “Two AI Markets” Theory
This new skepticism splits the AI world in two. Forget “good AI” and “bad AI.” Think of it as “Business-Changing AI” versus “AI-Flavored Window Dressing.”
Market 1: The “Window Dressing”
This is what most of the hype is about. You could call it “enhanced intelligence.” It involves using AI to do basic things. It’s cool, it’s flashy, but — and that’s a literally billion-dollar “but” — it’s not sophisticated enough to change your revenue or your margins.
This is the market where investors are getting nervous. Why? Because as one informed citizen might argue, while this market is very large, it may not be creating enough actual, hard-dollar value to justify the staggering cost of the large language models (LLMs) and their implementation.
Companies are spending billions, and what are they getting? A slightly better chatbot? A more poetic email auto-complete?
Market 2: The “Business Changer”
This is the other market. This is the sophisticated, heavy-lifting AI. This is the stuff that governments and massive enterprises use for things like completely overhauling a supply chain, finding new military targets, or fundamentally restructuring logistics.
This stuff absolutely changes revenue and margins. This is the “rational” part of the rational bubble.
The Real Problem: A Bad Trade
Here’s the trap: Everyone is paying Market 2 prices, but many are only getting Market 1 results.
The investor concern is no longer just “is this a bubble?” It’s “is this a bad business decision?”
Think about it. We’re watching companies pour billions of dollars into AI spending. But if that spending only produces “enhanced intelligence” for basic tasks, it’s not an investment; it’s an unjustifiable expense. It’s like buying a Formula 1 car just to drive it in a 20-mph school zone. The cost is astronomical, and the value? Negligible.
This refines our idea of the “coming tears.” The tears won’t just be from a bubble popping; they’ll be from CFOs looking at their balance sheets in two years and realizing they spent a fortune on AI-powered tech that didn’t move the needle one bit.
The revolution is real, but the accounting is, too. The most important question for any investor isn’t “Does this company use AI?” — it’s “Can they prove their AI is worth the price tag?”