The ‘smart money’ isn’t acting like we’re in a bubble, top economist says. The AI ballgame is in its ‘early innings’

Owen Lamont of Acadian argues that despite frothy valuations—S&P 500 briefly eclipsing 7,000 and a Shiller CAPE near 40—the market lacks the hallmark of a financial bubble: equity issuance, with U.S. firms instead repurchasing roughly $1 trillion of stock over the past year. He flags his Four Horsemen framework (overvaluation, bubble beliefs, issuance, inflows) and says a true bubble would be signaled by a surge in IPOs/insider selling; watch 2026 for potential triggers including high-profile listings (OpenAI rumored for Q4 2026, SpaceX) and heavy pipelines from firms like Blackstone and Goldman that could change investor positioning.

Analysis

Market structure: The current market is being propped up by corporate buybacks (~$1T last year) and concentrated AI-capex winners (NVDA, MSFT, ORCL) rather than fresh primary issuance; that favors cash-flow rich incumbents and exchanges/banks that monetize deal flow (BX, GS, NDAQ) if IPOs accelerate. Reduced float compresses supply and amplifies momentum, but it also raises liquidity fragility—a supply shock (large IPO wave) would increase float and could invert pricing power within 6–12 months. Expect price leadership to concentrate in semiconductors and hyperscalers, while late-stage private investors and high-cost capital providers face margin pressure if valuations mean-revert.

Risk assessment: Tail risks include an IPO-driven “smart-money exit” in H2–Q4 2026 (trigger: >$200bn primary issuance or >100 IPOs quarter), a regulatory bifurcation of AI winners/losers (antitrust or data-privacy fines), or a data-center overbuild that collapses GPU pricing (20–40% downside for chip-equipment-makers). Short-term (days–weeks) market moves will be headline-driven (OpenAI IPO rumors); medium-term (months) driven by issuance/flow data; long-term (quarters–years) driven by AI capex ROI and corporate earnings vs. capex write-offs. Hidden dependency: buyback-funded EPS supports mask true free-cash-flow economics—if buybacks slow 30% year-over-year, EPS could rebase materially.

Trade implications: Tactical opportunities: long IPO facilitators (BX, GS, NDAQ) 1–3% positions ahead of a possible 2026 deal wave, long MSFT (1–3%) for durable AI exposure, and hedge NVDA exposure via 3–6 month put spreads (10–20% OTM) rather than outright shorts to cap tail risk. Consider pair trades: long GS or BX vs short late-cycle software names with weak balance sheets; use options to sell covered calls on concentrated NVDA positions (1-month +20% OTM) and buy protection for 3–6 months if conviction is low. Reweight away from retail/meme beta and increase cash-ready allocation by 5–7% into Q3–Q4 2026.

Contrarian angles: Consensus mistakes: equating AI capex with a classic irrational bubble ignores that corporates are rational capex actors—issuance is the true bubble signal. The market may be underpricing the risk that an IPO mega-cycle (OpenAI, others) will reverse buyback-driven scarcity and create a 15–30% re-rating in growth multiples within 6–12 months. Historical parallel: 1999–2000 showed issuance precedes the terminal phase; absent issuance, momentum can persist longer but with asymmetric downside when supply returns. Unintended consequence: heavy IPO supply could stress market-making inventories, widening bid-ask spreads and option vols—favoring liquidity providers and disadvantaging levered long-only funds.

Key Decisions for Investors

• Establish a 2–3% long position in BX and GS within 2–4 weeks to front-run an IPO pipeline; trim if primary issuance exceeds $200bn over any rolling 3-month period or IPO count >100 in a quarter.
• Establish a 1–3% long position in MSFT (buy shares or 9–12 month 5–10% ITM call spreads) for durable AI exposure; hedge 25–50% of position with 3–6 month put spreads (10–15% OTM) if NVDA/MS market cap leadership extends further.
• Reduce outright NVDA long exposure by 25–50% from peak weight; replace with a 3–6 month collar: sell 1-month calls at +20% OTM and buy 3–6 month put spreads (10–20% OTM) to limit downside to defined range.
• Establish a 1–2% long position in NDAQ to capture deal flow fees; if OpenAI files S-1 or signals price range, add another 1% within 2 weeks of the filing.
• If corporate buybacks decline >30% YoY or S&P net primary issuance flips positive, reduce cyclical equity beta by 25% within 30 days and rotate into cash/short-duration Treasuries (T-bill yield pickup target +50–100bps vs current).