One of the biggest mistakes investors make with secular growth themes is worrying too early about how it all ends. With growth investing, the journey matters more than the destination. Yes, we need to watch for yellow flags, but history shows there are usually more false alarms than fatal ones. Bull markets and transformative themes are built on exactly that—climbing a wall of worry.
Make no mistake: AI is a spectacular boom that will become a bubble. But bubbles can run longer, get bigger, and become far more disruptive than most expect. Hedgeye CEO Keith McCullough flagged this in his “Early Look” morning note several days ago, and I’ve lived through it. As a junior analyst in the late 1990s, I watched the dot-com mania consume not just markets but culture itself. 1999 was bonkers. We aren’t there yet—but if history rhymes, it’s important to recognize how powerful the psychology can get.
Sentiment: Not Linear, Never Predictable
Positioning moves fast.
Just a month ago, hedge funds were positioned cautiously in AI baskets, according to Goldman Sachs’ prime brokerage data. Now, the data shows they’re very bullish again. That’s why, in the short term, AI could lag while capital chases the next shiny object—right now, that’s Small Caps (IWM).
5 Reasons Why This Cycle Is Fundamentally Stronger
1. Economics Still Rational – Capex is chasing real demand. Token growth is parabolic. SNOW’s revenue acceleration confirms enterprises are committing more deeply to AI projects. Even quirky apps (like GOOGL’s nano banana) saw off-the-charts usage. The only irrational move? Zuckerberg’s billion-dollar “engineer raids,” which signal Meta’s desperation. Otherwise, hyperscalers are funding capex with cash (boosted by OBBBA), without destroying FCF margins—Meta again the exception.
2. Buy the Shortage, Short the Abundance – The industry is still short on GPUs, high-bandwidth memory, flash, and—most critically—power (electricity, turbines, transformers, nuclear). This doesn’t end until lead times for these critical components shrink materially.
3. Cash-Driven, Not Credit-Driven (Yet) – So far, this boom has been largely funded by cash. Credit has only just started flowing into this space this year. Contrast that with dot-com, where the late stage was fueled by IPO capital and credit (the fiber boom). Rising credit costs killed that party. The bell ringer? Time Warner buying AOL in January 2000—a deal that never made economic sense.
4. NVDA’s True TAM is Underestimated – Analysts are laser-focused on AI capex, but that’s just one of three simultaneous S-curves. Blockchain and crypto are also compute-intensive and GPUs/accelerators are the compute architecture. What’s modeled today is only the tip of the iceberg
5. The Capex Supercycle – The chart below from Morgan Stanley makes it clear: the U.S. has already undergone two massive transformations in the past 30 years—Silicon and Shale. AI infrastructure could be the third. Don’t underestimate how big this could become.
Where We Are: Browser Wars & Fiber Buildout
Using the dot-com analogy, we’re in the early infrastructure phase. Enterprises are seeing utility (AI coding software), and consumers are seeing search and productivity benefits. However the birth of Amazon, eBay, and Priceline reshaped economic behavior. We haven’t had that moment yet in AI. Maybe an iOS-integrated AI assistant on over a billion devices in 2026 changes that—ushering in true consumer adoption of agentic AI.
Potential Catalysts Lining Up
- Accelerating cloud revenues in Q3
- More hyperscaler infrastructure deals
- Sovereign AI deals
- AAPL partnerships (Gemini, OAI, or both)
- 2026: the rise of SLMs (specialized language models)
- A possible OpenAI IPO – could this be the first trillion-dollar IPO?
All of these could extend the boom to bubble-like levels.
How It Ends: Psychology, Not Spreadsheets
Bubbles don’t die on valuation alone. They die on sentiment and signals. In 2000, Cisco and Sun Microsystems both peaked above 100x forward earnings. NVIDIA today trades around 35x. That’s not a forecast—it’s context. It shows just how far markets can stretch when psychology takes over.
Bottom Line: This will become a bubble, but we are not yet in the final stage of mania. Demand is real, shortages remain, and the fundamental underpinnings still make sense. The endgame will be shaped less by DCF models and more by signals, psychology, and—eventually—one last spectacular bell ringer.
