Wall Street is in a state of extreme fear over the "AI Bubble" burst. But Gene Munster has just presented an counterintuitive conclusion: it’s 1995 (the dawn of the internet), not 1999 (the eve of the crash).

This is not a gamble driven by hype, but a "Pascal’s Wager" for survival. Giants must spend money because the cost of not doing so is death.

Below is the underlying strategic logic and asset revaluation guide for the 2026 market.

1. Pascal’s Wager: Why Capex Must Surge by 50%

Many people don’t understand why Meta and Google are frantically buying GPUs. It’s not because they are confident AI will be immediately profitable, but based on a brutal game theory:

· Option A: Invest heavily in infrastructure. Risk: If AI fails, lose hundreds of billions in cash. Outcome: Stock price drops, but the company survives.

· Option B: Stay conservative, cut expenses. Risk: If AI succeeds and you lack computing power. Outcome: Be eliminated by the times, removed from the list of giants.

Choose the lesser of two evils. For hyperscale cloud companies (Hyperscalers), over-investment is rational, under-investment is deadly. That’s why capital expenditure in 2026 will still irrationally surge.

2. The Shift of Value Anchors: “Thinking” > “Connecting”

Looking back at the 2000 internet bubble, fiber optic companies laid too much network, causing bandwidth prices to plummet toward zero. Transmission became a commodity.

But the logic in 2026 is different. We are creating “thinking” through AI. Unlike bandwidth, “intelligence” does not rapidly depreciate. Companies with proprietary infrastructure (like Google TPU) that can produce “intelligence” at low cost will hold pricing power.

This is also why Google is predicted to be the best-performing stock in 2026 — it’s not just about search, but about controlling the “thinking” production line.

3. The Curse of the Law of Large Numbers: The Turnaround for Small Caps

“The Big Seven” (Mag 7) are strong, but mathematical laws are unavoidable. To double the market cap of a $3 trillion company requires astronomical energy.

Funds will start overflowing. Alpha returns in 2026 will appear in “small-cap tech stocks” — but in the AI era, the definition of market cap is being reshaped. Here, “small caps” refer to companies with a market value below $500 billion.

Under the dual effects of a rate-cut cycle and AI technology sinking, these “second-tier giants” will outpace the Mag 7.

4. Apple’s Hidden Fortress: The Rebirth of Siri

The market underestimates Apple. Because everyone is focused on cloud models, they overlook “edge privacy.”

Apple has data even OpenAI cannot touch: your local life. It’s predicted that before April 2026, the new Siri will, by integrating device-side data, leap from “voice assistant” to “personal agent.” This moat is a physical barrier that pure software companies cannot cross.

5. Physical AI: Tesla’s New Coordinates

Viewing Tesla as just an automaker is outdated since 2020. With Waymo already leading 50:2 in autonomous driving, Tesla’s core value lies in its massive installed base.

It is the carrier of “physical AI.” Whether it’s Robotaxi or humanoid robots, Tesla is the only entity capable of embedding AI brains into physical shells and mass-producing them. The valuation restructuring of this part has not yet been fully reflected in the stock price.

6. The Truth of the Bubble: Rational Prosperity

The current rise is supported by fundamentals (data center demand, earnings growth), not just market sentiment ratios. When everyone doubts a “bubble,” it usually doesn’t burst. The real danger signal is when taxi drivers start recommending GPU stocks to you. Clearly, we are still 2-3 years away from that stage.

Deep insights:

I. Energy is the ultimate hard currency. As AI compute demand explodes exponentially, the bottleneck will shift from “chips” to “electricity.” Any company providing stable, clean base load power (nuclear, even natural gas) is essentially mining “shovels” for AI. Future tech giants may also be energy giants.

II. The “liquidity trap” in private markets. The delayed IPOs of OpenAI and Databricks reveal the failure of the IPO mechanism. Primary market funds are so abundant that retail investors are not needed to take on IPO risk, leading secondary market investors (ordinary shareholders) to only enter when these companies become trillion-dollar behemoths and growth slows. This is the biggest structural injustice facing retail investors in the AI era.

III. The “reflexivity” of infrastructure. The current AI building boom itself is creating demand. This “investment-output” flywheel lasted 4 years from 1995 to 1999. If you get off now, you will miss not only the gains but also the largest productivity leap in human history. Don’t be the person who sold Amazon in 1996.
#AI泡沫 #Nasdaq #Investment Logic