Should You Fear The AI Bubble?

2025's tech and AI rally has prompted growing bubble talk after Alphabet CEO Sundar Pichai acknowledged elements of irrationality and Nvidia—despite beating Q3 estimates and giving strong Q4 guidance—saw the market retrench as payroll data raised odds that a December Fed cut is less likely and Nasdaq slipped about 2%. The author compares today to past bubbles: unlike the dot‑com era, most AI leaders (Nvidia, AMD, Oracle) have real products and growing revenues, but valuation froth exists in smaller or unprofitable names (Palantir traded near ~110x revenues in November; CoreWeave is off >50% from its highs), while aggregate metrics remain elevated (Shiller PE ~39x; Buffett Indicator ~230%). Conclusion: a broad-market dot‑com style collapse is deemed unlikely, but downside risk is concentrated in high‑multiple and unprofitable AI/tech stocks; the author is not adding to QQQ exposure and favors lower‑risk allocations in cheaper sectors (healthcare, financials, energy, infrastructure) while recognizing asymmetric upside for concentrated AI bets.

Analysis

Short-term market chatter about an "AI bubble" intensified after Alphabet CEO Sundar Pichai acknowledged elements of irrationality in the space and NVIDIA reported a Q3 beat with strong Q4 guidance; despite the beat the market sold off on Thursday amid stronger payrolls that reduced the odds of a December Fed cut and contributed to a roughly 2% pullback in the Nasdaq (QQQ). Historical comparison to the dot‑com era shows differences: while the Nasdaq soared ~700% and the S&P ~200% in the 1995–2000 run, today's leaders (NVIDIA, AMD, Oracle, Microsoft, Alphabet) mostly have real products, revenues and order books rather than purely speculative business plans. Valuation stress is concentrated: Palantir traded near ~110x this year's revenues in November and CoreWeave is off >50% from its highs, while aggregate metrics remain elevated with a Shiller PE of ~39x and a Buffett Indicator around 230%, reflecting outsized weight of high‑quality large caps. The author judges a broad dot‑com style collapse unlikely but flags elevated downside for high‑multiple and unprofitable AI names, prefers not to add to QQQ, and recommends redeploying risk into cheaper, out‑of‑favor sectors while monitoring earnings, order trends and monetary policy signals.

Key Decisions for Investors

• Do not add to broad tech/QQQ exposure right now; consider trimming or avoiding high‑multiple, unprofitable AI names such as recent CoreWeave-like cases and very stretched names like Palantir at ~110x revenues
• Increase allocations to cheaper, out-of-favor sectors highlighted by the author — healthcare, financials, energy and infrastructure — to lower portfolio valuation risk
• If retaining exposure to AI leaders (NVDA, AMD, MSFT, GOOG), keep position sizes disciplined and monitor quarterly revenue, order trends and guidance rather than chasing further multiple expansion
• For concentrated AI bets use position sizing limits or hedges and watch macro indicators (payrolls, Fed guidance) and valuation gauges (Shiller PE, Buffett Indicator) as catalysts that could widen downside risk