He Predicted the Dot-com Bubble Burst. Now He's Saying SpaceX Could be a Fresh Warning Sign.

Jeremy Grantham (GMO) warns the AI-driven rally could be nearing its end, citing the massive IPO of Space Exploration Technologies that raised $75B (about $86B including underwriters’ overallotment) as a potential bubble signal. He notes the 1990s pattern where large internet stock IPOs appeared near the top, but also concedes timing is uncertain. The takeaway for portfolios is to reduce material AI exposure/tilt toward value or more resilient sectors rather than panic-sell due to limited timing certainty.

Analysis

The signal here is less about one issuer and more about capital formation becoming a liquidity event for a crowded theme. Very large, story-driven listings typically pull marginal dollars away from the highest-duration names first, which means the first pressure point is usually not the megacap AI leaders but the speculative second-derivative trades around them: unprofitable software, recent IPOs, and theme baskets with weak cash generation. If that capital rotation sticks, the market gets narrower before it gets weaker.

Near term, this is more a positioning risk than a fundamental earnings shock. NVDA and other profitable AI bellwethers can keep outperforming for weeks if hyperscaler capex stays firm, but their multiple support becomes more fragile once investors have a new venue for AI optionality and a large amount of fresh paper hits the market. The names most exposed are the ones that need continued sentiment expansion to justify valuation: ARKK-style growth, pre-profit software, and any SPAC/IPO complex that trades on scarcity rather than cash flow.

The contrarian view is that timing calls from bubble watchers are usually early, sometimes by quarters. A mega-IPO can mark peak enthusiasm, but it can also simply confirm that the market still has ample risk appetite. What would falsify the bearish setup is continued upward revisions to 12-month AI capex, stable credit spreads, and semis breadth holding even if post-IPO performance is weak.

Key Decisions for Investors

• Use XLU/XLP as the defensive leg against high-duration growth over the next 1-3 months; pair long XLU or XLP vs short ARKK if breadth starts to roll. Risk/reward: modest carry, but strong convexity if the market rotates out of speculative AI.
• Buy a 2-3 month QQQ put spread as a portfolio hedge rather than an outright short. Structure for a 5-8% index pullback; this is cleaner than shorting NVDA directly because it isolates valuation compression risk without fighting earnings momentum.
• Do not fade NVDA outright on this headline; if you need AI exposure, express it as a quality-vs-narrative pair: long NVDA, short ARKK or an unprofitable software basket. Time horizon: 1-3 months, with the short leg intended to capture multiple compression if risk appetite cools.
• Treat SPCX as a sentiment short only if it trades into aftermarket/IPO strength: fade the first 1-2 weeks of price discovery or wait for secondary supply/lockup timing. The trade is about liquidity exhaustion, not fundamentals.
• Set an alert on hyperscaler capex commentary and semiconductor breadth; if those stay firm, the bubble narrative stays premature. If breadth weakens while IPO issuance stays hot, reduce cyclically exposed growth by 20-30%.