Navigating the AI Boom: Bubble is NOT a Four-Letter Word

The author argues that the AI boom may still have room to run, citing a Nasdaq 100 trailing P/E of about 29.5x versus more than 100x for many leaders at the March 2000 peak, and NVIDIA’s stated P/E of 34.67x. The article also points to 251 IPOs in 2026 versus 446 in 2000, SanDisk trading 47% above its 200-day moving average, and AAII bearish sentiment exceeding bullish sentiment as signs that euphoria is limited. It acknowledges that historical bull markets eventually end and losses can be severe, but says current AI spending is supported by revenue and argues investors should not avoid the opportunity solely because it is called a bubble.

Analysis

The article’s bullish case is weaker as a timing signal than as a reminder that AI adoption can create real value. Productivity gains do not establish that infrastructure providers earn attractive returns: bargaining power may migrate to cloud platforms and enterprise users, while hardware, memory and data-center capacity face price competition and periodic overbuild. This is the key second-order risk for NVDA and SNDK; demand growth alone is insufficient if customer utilization, pricing or returns on invested capital disappoint.

The valuation comparison is not decisive: an index multiple and a single company’s multiple do not control for growth, earnings quality or cyclicality. Nor does bearish retail sentiment reliably mark an early phase; it can coexist with crowded institutional ownership and narrow leadership. Treat the article’s figures and product anecdotes as claims to verify, not evidence of broad monetization.

Near term (days), the piece itself is unlikely to change earnings estimates. Over 1–3 months, watch hyperscaler capex guidance against disclosed AI revenue/productivity monetization, and semiconductor order revisions. Over 6–18 months, capex returns and capacity utilization should determine whether AI leaders sustain multiples. A slowdown in orders or weaker returns could reprice the whole supply chain even if adoption continues. The quantum-computing promotion is not an investable catalyst without commercial milestones.

Key Decisions for Investors

• Avoid buying a broad AI basket solely on the article’s sentiment and historical analogies. For incremental exposure, consider a small, defined-risk NVDA call spread around the next earnings catalyst; enter only if forward demand commentary and estimates remain firm. Loss is limited to premium; exit or reassess on order/guidance deterioration or a sharp post-event reversal.
• Prefer exposure to monetization over an indiscriminate infrastructure chase: monitor META, MSFT, GOOG and AMZN for evidence that AI features support engagement, retention or cloud demand. Do not infer revenue impact from demonstrations; require measurable disclosures or guidance revisions before adding.
• Use a 1–3 month alert rather than a short recommendation: if hyperscaler capex rises while AI-related revenue, utilization or supplier orders fail to keep pace, reduce semiconductor and storage exposure, including NVDA and SNDK. Falsification of this caution would be sustained order growth alongside improving disclosed returns on investment.
• Do not treat bearish AAII sentiment as a standalone buy signal. Track market breadth, concentration and earnings revisions; broadening participation with upward estimate revisions would support the bull case, while deteriorating breadth and downward revisions would invalidate it.