Stop Asking When the AI Bubble Will Burst — Here's the Real Story

"When will the AI bubble burst?" — more and more people are asking this question, with a tone of excitement, anxiety, or both. Recently, a tech media editor wrote plainly that even his own colleagues keep pressing him on this, and his answer was blunt: if I could predict when the bubble will burst, I'd have already gone all-in with maximum leverage, achieved financial freedom, and then written this article.

This half-joking response cuts right to the core dilemma of the current AI bubble debate: nobody has a crystal ball. But that doesn't mean the discussion is worthless — what matters is how we frame the question.

Predicting the Timing of a Bubble Is Fundamentally Impossible

Market bubbles never burst due to a single factor. They're typically triggered by an interplay of interest rate conditions, investor sentiment, the pace of technology adoption, regulatory shifts, and more — and these variables feed back into each other in complex loops. Even seasoned investors who have lived through multiple bubble cycles rarely nail the turning point with precision.

The original author's stance is clear: rather than chase an unanswerable date, put energy into more actionable questions. This isn't avoidance — it's an honest acknowledgment of the limits of prediction. Historically, most who claimed to time the market precisely were eventually proven wrong.

Bubble in economics typically refers to a state where asset prices deviate significantly from their intrinsic value, driven mainly by speculative expectations rather than fundamentals. Identifying a bubble is itself contentious: ex ante, it's hard to distinguish "irrational exuberance" from "reasonable pricing of new technology prospects" — confirmation often only comes in hindsight. As a result, academia tends to use valuation metrics (such as historical percentiles of P/E and P/S ratios) and sentiment indicators (such as retail trading share and margin balances) to gauge the degree of risk accumulation, rather than pinpointing a burst date.

Market timing refers to attempting to predict market turning points and adjust positions accordingly. A large body of empirical research shows that over the long run, timing strategies have unstable success rates; the few success stories are hard to replicate, and trading costs and taxes erode returns. This is the deeper reason the original article stresses that timing cannot be predicted.

What Questions Should We Actually Be Asking

Since "when will it burst" is hard to answer, what questions are both answerable and meaningful?

What Growth Expectations Are Baked Into Current Valuations

The essence of a bubble is the gap between price and value. Rather than guessing the turning point, break down how much revenue growth and margin expansion is implied in current AI-related asset pricing. If those assumptions require near-perfect execution to deliver, the risk is plainly there. This is a question that can be approximated through financial models and public data.

Is the Gap Between Technology Adoption and Commercial Returns Widening or Narrowing

AI capability is iterating at a visible pace, but there's still a conversion cycle between capability and sustainable commercial returns. Watching enterprise willingness to pay, renewal rates, and whether AI features are truly embedded in core workflows reflects the real direction of fundamentals far better than staring at stock price swings.

Is the Payback Period on Capital Expenditure Being Underestimated

The depreciation burden and payback period of large-scale compute investment are key dimensions for assessing the health of AI investment. When capex growth consistently outpaces revenue growth, the question to ask is: what demand assumptions underpin this pace of investment?

Capital expenditure (CapEx) refers to spending on purchasing or upgrading long-term assets such as data centers, GPU clusters, and networking equipment. Unlike day-to-day operating expenses, capex is amortized through depreciation over multiple years. In AI infrastructure investment, hardware like GPUs iterates quickly and may have an economic life shorter than its accounting depreciation schedule — if actual useful life falls short, book depreciation may understate true cost pressure. This is an easily overlooked detail when assessing the health of AI investment.

Why "Bubble Prophecies" Themselves Deserve Caution

An obsession with the bubble bursting tends to breed two cognitive biases. One is exiting too early and missing structural opportunities in a long-term technology trend; the other is trading too frequently, getting dragged around by short-term noise and racking up friction costs.

The original author's hypothetical scenario of "betting with leverage" actually illustrates a simple truth: genuine certainties don't present themselves in a way that everyone can discuss openly. When a question keeps being asked but never reaches consensus, it's better treated as a background variable for risk management than as a trading signal.

Two cognitive biases are most relevant here: first, confirmation bias — people tend to seek information that supports their existing views, so bears keep amplifying negative signals; second, loss aversion — sensitivity to losses exceeds the pleasure from equivalent gains, which can prompt premature selling of profitable assets or frequent position adjustments. Understanding these biases helps pull the "bubble" discussion from the emotional level back to the level of decision discipline.

A Practical Mindset for Followers

Rather than waiting for a precise date that will never come, build a framework for dealing with uncertainty: stay engaged with AI's long-term trend, stay clear-eyed about short-term valuations, and stay honest about your own risk tolerance. Whether the bubble bursts, and when, will ultimately be answered by the market — not announced in advance by any commentator.

So the next time you're about to ask "when will the AI bubble burst," ask yourself first: what I really want to know — is it that date, or is it how to make more robust decisions in this uncertain environment?

One-line Summary

Rather than asking when the AI bubble will burst, focus on valuation assumptions and the gap with fundamentals.

Paragraph Summary

The article responds to the popular question "when will the AI bubble burst," pointing out that precisely predicting a bubble's turning point is fundamentally impossible, since bubbles are triggered by an interplay of interest rates, sentiment, technology adoption, regulation, and other variables. Using the hypothetical "if I could predict it, I'd bet with leverage," the author shows that genuine certainties don't exist in a form everyone discusses. The article suggests redirecting attention to answerable questions: what growth expectations are baked into current valuations, whether the gap between technology adoption and commercial returns is widening or narrowing, and whether the payback period on capex is being underestimated. It also warns that obsessing over the bubble bursting can lead to two biases — exiting too early or trading too frequently — and that a more robust approach is to build a framework for uncertainty rather than wait for a precise date.