What is a bubble? Why finance teams are watching AI risk right now
AI-linked equities have ripped higher this year. Some desks see froth. Others say the move reflects durable earnings capacity and real adoption. Your job isn't to predict the headline - it's to price the risk and protect capital.
What is a stock market bubble?
Working definition: A bubble is a sharp run-up in asset prices without a matching increase in business value. Sentiment outruns cash flows, and buyers keep bidding because they expect the next buyer to pay more.
Once optimism fades, the crowd rushes for the exit. Selling feeds on itself, liquidity thins, and prices reset fast.
History offers a clear case: the late-90s dot-com period. Investors chased anything with a website; when the air came out in 2000, the Nasdaq Composite fell by nearly 80% over the next two years.
Useful references
How to spot a bubble before it bites
- Price vs. information mismatch: Shares keep rising after earnings misses, weaker guidance, or downgrades.
- Narrative over numbers: The story is "this changes everything," while unit economics, margins, or retention don't confirm it.
- New money dominance: A wave of novice participation and social proof, with claims that experienced investors "don't get it."
- Metric drift: Emphasis on adjusted or novel KPIs while GAAP profitability and cash generation lag.
- Primary issuance: More secondaries, converts, or IPOs as companies capitalize on elevated valuations.
- Insider behavior: Rising insider selling and shrinking buybacks.
What a bubble does to investors and companies
- Late-entry losses: Those who buy after the run-up often absorb the drawdown.
- Spillovers: Sharp repricing can hit broader demand, push the economy into a downturn, and depress portfolios and hiring.
- Corporate strain: Impairments, tighter covenants, working capital stress, and tougher refinancing terms.
Practical checklist for AI exposure (public markets)
- Valuation discipline: Tie enterprise value to credible revenue, margin, and free cash flow paths. Stress test with higher discount rates and lower terminal growth.
- Position sizing: Cap single-name risk; avoid correlation clusters across AI infra, enablers, and applications.
- Hedging: Consider protective puts on sector ETFs, collars on concentrated positions, or pairs trades (long quality, short a basket of weaker peers).
- Liquidity: Monitor average daily dollar volume; plan exit routes before volatility spikes.
Practical checklist for AI capex and M&A (corporates)
- Unit economics first: Track gross margin after compute, data, and serving costs. Map customer payback period and cohort retention.
- ROI gates: Stage-gate spending by milestone (pilot → limited rollout → scaled deployment). Require clear owners, KPIs, and sunset criteria.
- Cash discipline: Separate experiments from production budgets. Avoid long, non-cancellable commitments without price-to-performance step-downs.
- Operational proof: Validate productivity claims with baseline vs. post-implementation metrics (cycle time, defect rate, throughput).
Key metrics to scrutinize in AI-linked names
- Revenue quality: New logo mix, expansion vs. new, and NRR. Look for durable, not one-off, deals.
- Margin trajectory: Gross margin vs. compute and data costs; opex growth vs. revenue growth.
- Cash conversion: Free cash flow after capex; working capital intensity; SBC as a % of revenue.
- Customer economics: CAC, LTV, sales cycles, and payback period stability as competition rises.
If the AI bubble pops: a fast-response plan
- Scenario pack: Pre-build -20%/-40%/-60% cases with triggers for hedging, de-risking, and expense actions.
- Funding runway: Secure credit capacity before you need it; revisit covenants and collateral.
- Portfolio triage: Prioritize high-IRR projects; pause low-visibility bets; renegotiate vendor terms tied to compute and data.
- Communication: Align leadership on the plan to reduce rumor risk and retain key talent and customers.
Bottom line
Bubbles are stories priced as certainty. As a finance lead, anchor decisions to cash flows, unit economics, and risk limits. If the story holds, you compound. If it doesn't, you survive to compound later.
Helpful resources
Your membership also unlocks:
