Goldman Sachs: AI Positioning Highly Concentrated, Index Calm Masks Underlying Fragility in US Equities
Taylor Wilson
The S&P 500 posted seven straight up-days and nine consecutive winning weeks, but Goldman trader Lee Coppersmith warns that highly concentrated AI bets and elevated leverage are creating fragility beneath the index-level calm.
The index looks steady — so what's the problem?
The S&P 500 closed higher for 7 consecutive sessions last week, extending its winning streak to 9 straight weeks. The Nasdaq 100 gained roughly 10% in May alone, rebounding 32% from its March low.
Yet Goldman's Coppersmith flags a widening divergence between index-level calm and violent rotation at the factor, positioning, and single-stock level.
This means → the "stability" investors see is an index-surface phenomenon. Underneath, individual stocks and sectors are churning hard — the index is a still lake with fierce undercurrents.
Which direction are investors betting on?
Average single-stock put/call skew (skew — a measure of whether investors are paying more for downside or upside protection) on the S&P 500 has fallen to the lowest level on record.
This means → investors are increasingly paying for upside acceleration, not downside hedges — even as realized volatility keeps rising.
Goldman's TMT momentum pair trade has seen over 25 single-day swings of ±5% this year, versus just 6 for all of last year. In plain terms = everyone says "the market is calm," but positioning is aggressively betting AI keeps running.
Software rallied 8% — who's winning inside the sector?
The software sector gained roughly 8% last week, driven by better-than-expected earnings from OKTA and SNOW and optimism around AI monetization.
But Goldman's trading desk notes a clear divergence within the sector: data-infrastructure and security sub-sectors — viewed as "infrastructure winners" — keep strengthening, while more traditional SaaS models face pressure.
This reflects the market running a finer screen on the AI beneficiary chain — not every software company will capture the AI dividend.
Why is hedging getting harder?
S&P 500 implied correlation (implied correlation — a gauge of how much individual stocks move in sync) sits at historic lows, pricing in an exceptionally benign environment.
Historically, macro shocks trigger sharp correlation spikes. But Coppersmith concedes that over the past few years this type of hedge has been extremely painful to hold, because the market pattern has been the exact opposite — narrow leadership, wild single-stock moves, and an orderly index.
In plain terms = the market structure is a taut string: one systemic shock and all stocks snap into synchronized decline — yet almost nobody is insuring against that scenario right now.
Up 85% — is it all speculation?
The S&P 500 has gained roughly 85% since ChatGPT's launch, but Coppersmith stresses this is not pure valuation froth.
Goldman's US strategy team forecast 2023 S&P 500 EPS at roughly $224 before AI; the current 2026 EPS forecast is $340 — about 52% above the pre-AI forward estimate. Roughly two-thirds of the index's gain can be explained by earnings growth.
This means → this bull market has real earnings support, but the remaining third is valuation expansion and sentiment premium — where AI value ultimately accrues in the chain remains unclear, and that very uncertainty is paradoxically sustaining the breadth of global AI spending enthusiasm.
Content is for reference only, not financial advice.