The following summarises a Bloomberg Opinion column by Nir Kaissar, founder of asset management firm Unison Advisors. It reflects the author’s personal views and argues a position much of the market currently rejects.
- Kaissar argues this is among the most distrusted rallies he has followed, and that the absence of animal spirits is the strongest evidence against a bubble. American Association of Individual Investors surveys showed bulls outnumbering bears by more than 20 percentage points on rolling three-month averages in the late 1990s; bears now lead bulls by 8 points, a swing of nearly 30 points.
- His valuation measure is an unconventional one worth understanding: forward price-earnings divided by operating margin. That adjusts for a more profitable index than historical versions, on the view that higher margins justify a higher multiple, and on that basis he finds the market modestly above average rather than extended.
- The same column supplies the counterargument. AI-related companies account for 40% to 60% of US market value depending on definition, and Citadel Securities estimates they produced nearly 70% of the S and P 500’s third quarter earnings growth. In 1999, three of the five largest US stocks were old-economy businesses.
- His forecast is specific and testable. Profit growth will cool from the current 25%, a market pullback will signal the downshift, earnings growth will fall to high single digits then turn negative for a stretch, speculation and fraud will be exposed, and absent a crisis the result should be a garden-variety bear market rather than worse.
What Happened?
Kaissar describes himself as an AI optimist who expects the benefits to outweigh the costs and the technology to enhance work and create jobs, while acknowledging the backdrop is less hospitable than the 1990s, with regional wars, widening inequality, uncontrolled deficits and a divided Congress. He notes the AI boom feels to many people more like a threat than progress, that graduates are struggling to find work as AI encroaches on entry-level white-collar roles, and that a growing number of companies are withdrawing or postponing listings including Anthropic and OpenAI. He closes by observing that the more people see a bubble, the less likely there is to be one.
Why It Matters?
Sentiment and concentration measure different things, and the column’s own figures describe a market that is both unloved and heavily concentrated. A 30-point swing in the bull-bear spread since 1999 is genuine evidence that positioning is cautious, which historically limits how far a decline runs because fewer investors are over-committed. It says nothing about what happens if one theme producing 70% of earnings growth falters. Those two facts coexist comfortably, and readers weighing the argument should hold both rather than choosing between them. The concentration risk is structural; the sentiment reading is about how crowded the exit is. The margin-adjusted valuation measure deserves attention because it is not standard and it changes the conclusion. Comparing today’s multiple to history without adjusting for profitability treats a more profitable index as equivalent to a less profitable one, and the adjustment is defensible. It also depends entirely on current margins being sustainable, and margins driven by a capital expenditure cycle are precisely the ones that mean-revert. Kaissar’s own forecast of earnings growth falling to high single digits and then turning negative is a prediction that the denominator of his valuation measure deteriorates. The contrary evidence from within the market is worth placing alongside. Goldman Sachs traders wrote last week that AI capital expenditure has likely peaked and that the debate has shifted to financing, sustainability and returns on capital rather than near-term earnings. A Federal Reserve president has warned the buildout may be overheating the economy, and BlackRock’s chief executive has said slowing it could damage the economy rather than just the market. Those are not bubble calls, and they describe the same transition Kaissar predicts, which suggests the disagreement is about severity rather than direction.
What Next?
Earnings growth is the variable Kaissar himself identifies, currently running at 25% and expected by him to cool to high single digits before turning negative. Watch the share of index earnings growth attributable to AI-related companies, since nearly 70% is the concentration figure that determines what a stall would cost. Sentiment surveys are the other half of his argument and a shift toward bullishness would weaken it on his own terms. Delayed listings, including those he names, are the market’s revealed judgement on appetite for new AI exposure.
Affected Tickers and Coins: SPY
Source: Bloomberg Opinion
