The math is starting to break.
Wall Street is finally looking at the denominator. For two years, the narrative of infinite scaling supported valuations that defied gravity. Now, the bill is coming due. Goldman Sachs Global Banking and Markets has shifted its tone from cautious optimism to structural skepticism. Shawn Tuteja, a key architect in the firm’s volatility trading desk, is pointing toward a specific fracture in the market: the custom basket. These concentrated clusters of AI stocks have become the most crowded trades in financial history. When everyone exits through the same narrow door, the price action turns violent.
The premium is evaporating. Investors are no longer paying for potential. They are demanding realized earnings per share. According to recent data from Bloomberg, the spread between AI-linked equities and the broader S&P 500 has reached a five year high. This divergence is not a sign of strength. It is a sign of isolation. The market is bifurcating into the AI ‘haves’ and the legacy ‘have-nots.’ This concentration creates a feedback loop where volatility in a single semiconductor name can liquidate an entire thematic ETF. We are seeing the mechanics of a liquidity trap forming in real time.
The Architecture of the AI Volatility Skew
Implied volatility is lying to you. While the VIX remains suppressed near historical lows, the internal correlation of AI baskets is screaming. Traders are paying massive premiums for out-of-the-money put options. This ‘skew’ suggests that while the surface looks calm, the professional money is terrified of a tail risk event. Tuteja’s analysis of custom baskets suggests that the ‘gamma’ profile of these stocks has become dangerously convex. A 3 percent move in the underlying index now triggers a 7 percent move in the derivative layers. This is not organic growth. This is a leveraged bet on a single outcome.
Capital expenditure is the new metric of doom. The ‘Magnificent Seven’ have committed over 200 billion dollars to AI infrastructure in the last fiscal cycle. However, the return on invested capital (ROIC) remains opaque. Software companies are struggling to monetize the massive compute costs they are incurring. Per reports from Reuters, enterprise adoption of generative AI tools has slowed as CTOs demand proof of productivity gains. If the revenue does not materialize by the next quarterly cycle, the CAPEX cycle will collapse. That collapse will start with the hardware providers and ripple through the entire tech ecosystem.
Visualizing the May Market Concentration
AI Basket Concentration vs Market Volatility (May 19)
The Custom Basket Liquidation Risk
ETFs are the primary transmission mechanism for this risk. When a retail investor buys an AI-themed ETF, the underlying algorithm buys a fixed ratio of stocks regardless of their individual valuations. This creates ‘forced buying’ on the way up and ‘forced selling’ on the way down. Goldman Sachs monitors these custom baskets because they represent the purest form of market sentiment. Currently, the delta between the basket’s price and its fundamental value is at a record wide. The market is ignoring the balance sheet in favor of the momentum trade.
| Asset Class | Forward P/E (May 2026) | 30-Day Volatility | Institutional Ownership |
|---|---|---|---|
| GS AI Custom Basket | 42.5x | 34.2% | 88.4% |
| S&P 500 Tech Index | 29.1x | 18.5% | 72.1% |
| Broad Market (Equal Weight) | 17.8x | 12.4% | 64.5% |
Institutional ownership in these baskets has crossed the 85 percent threshold. This is a danger zone. When institutional funds rebalance at the end of the quarter, they look for liquidity. The AI sector is currently the only place where large blocks can be moved without immediate slippage. This makes it the ‘piggy bank’ for the rest of the portfolio. If a hedge fund takes a loss in commodities or fixed income, they sell their AI winners to cover the margin. This cross-asset contagion is exactly what Tuteja is warning about in the volatility trading space. The rally is no longer about technology. It is about collateral management.
The Derivative Overhang
The options market is now larger than the underlying cash market for several key AI components. This is the tail wagging the dog. Market makers are forced to hedge their exposure by buying or selling the underlying stock as prices move. In a low-volatility environment, this provides stability. In a high-volatility environment, it creates a ‘gamma flip’ where market makers become sellers as the price drops, accelerating the decline. This technical setup is identical to the conditions seen before the 2000 dot-com crash and the 2021 growth stock correction.
We are currently observing a massive build-up in short-dated call options. Retail traders are gambling on weekly earnings ‘beats’ to sustain the momentum. However, the market has already priced in perfection. A ‘beat and raise’ is no longer enough to move the needle. Stocks are now falling on positive news because the ‘whisper numbers’ are significantly higher than official guidance. This is the exhaustion phase of the cycle. The narrative has shifted from ‘how much will they make’ to ‘how much did they overspend.’
The next critical data point arrives on June 4, 2026, with the release of the revised enterprise software spending forecast. If that number shows a contraction in AI budget allocations, the volatility skew we are seeing today will transform into a full-scale liquidation. Watch the 20-day moving average on the GS AI Basket. A break below that level will trigger the automated selling programs that now govern 70 percent of daily volume.
