ROI-AI tremor exposes market's volatility fault lines: Mike Dolan

The AI boom has kept market fears at bay ​for more than two years. October suggests that shield is beginning to crack. Last week's turbulence in the sector revealed investors preparing ​for a world where tech euphoria no longer drowns out all other macroeconomic signals.

That doesn't ‌spell ​the end of the AI story by a long shot, but it does signal greater discrimination within the sector as blistering growth rates crest, development matures, and the trade-off between computing costs and revenue reaches a more sustainable equilibrium. But a disturbance in one part of the increasingly intertwined, cross-financed ecosystem can quickly spread across broad indexes and other asset classes.

The most glaring evidence of that came last week, when government bonds suddenly broke ‌from spiraling oil prices, interest rate hike speculation and debt worries. They found a huge bid from investors seeking protection from the economic fallout of an AI slowdown or accident. While AI-flattered earnings growth has insulated US equity markets from surging borrowing costs all year, that protection may have reached its logical limit.

According to Societe Generale strategists, the scale of the shift can be captured by a rare period of correlation between equity prices and bond yields that reached its most extreme level since the late 1990s before reversing. They looked at the three-month correlation between S&P 500 daily log returns and daily ‌changes in the 10-year Treasury yield, and found that it reached its most negative reading in 30 years earlier this month, even with equities and bond yields both hitting two-decade highs.

The high-water mark of the AI trade, perhaps? "This increasingly negative sensitivity to rates has left the ‌market in a fragile equilibrium, with any further rise in yields becoming an incremental headwind to equities," the SocGen team wrote.

SocGen also said that implied equity index volatility, as measured by the Cboe Volatility Index (VIX), stirred from its slumber amid a "breakdown in volatility suppression." Whatever's rumbled through the month, it's confounded many in the hedge fund world. Goldman Sachs monitoring showed that systematic long-short hedge funds, whose algorithms ride market trends until they peter out, suffered their worst week since 2024, with a 2% decline.

UNDER THE BONNET To understand what is happening, look at the VIX volatility index and its sub-indexes — what they say about market risk and how speculative traders have used them.

Even though the VIX popped higher over the ⁠past couple of weeks as ​stocks were rattled by AI worries and bonds surged, it remains half asleep ⁠in the bigger picture, at about half its level shortly after the Iran war started eight months ago and below its historic average. Because the VIX is used as a key variable to signal trouble in the wider equity world and within multi-asset portfolios, regulators and central banks worry that it's understating the level of risk building.

They cite the "volatility ⁠suppression" trades that SocGen highlighted. Done with leverage and high volumes, these lucrative trades typically involve selling options or "shorting" implied-volatility indexes to profit from the usually positive premium of implied over realized volatility.

Not unlike hedge funds' so-called "basis trades" in Treasury markets, they are not directional bets on underlying prices, but ways to arbitrage persistent price gaps. One strategy ​bets on rising volatility in individual stocks while selling options on the index, relying on the idea that individual stocks may fluctuate sharply while evening out at the index level.

The problem is twofold and has emerged many times before. First, these work like a ⁠dream until unforeseen events trigger a dramatic unwind of the volatility trades themselves, sending index volatility far above what the event may justify. Because index volatility gauges are used in risk models across global asset markets, such an unwind can transmit shockwaves everywhere.

While turning points are difficult to predict, the current setup risks compounding shocks just as AI's earnings momentum begins to hiccup. Morgan ⁠Stanley's ​latest take on the AI trade this weekend remained upbeat about the boom, estimating another 50% increase in overall AI investment spending in 2027. But it also said the rising cost of capital would affect more selective and highly leveraged parts of the AI network, and more volatility may ensue.

"As the AI capital cycle matures, credit quality, collateral, and structure will matter more, not less," it said. "The capital will be there, but increasingly on different terms for different borrowers." Barclays credit analysts also insist that AI is too big now to be treated as a single corporate ⁠debt segment, with performance gaps among AI-linked borrowers widening rapidly.

Even though overall AI-related debt is underperforming investment-grade credit indexes this year by about 35 basis points, that masks a near-tripling of dispersion among AI names to 65 bps - the highest in recent history. Even excluding outliers like Oracle, ⁠dispersion has risen 60%. Whether all this churn under the hood still justifies ⁠low index volatility is a huge question, not least with non-AI risks mounting into year-end from a busy election calendar, the Iran conflict and the energy outlook.

There is more than enough on the calendar to test investor nerves. Markets may simply be missing the warning lights. (The opinions expressed here are those of Mike Dolan, a columnist for Reuters.)

Enjoying this column? Check out Reuters Open Interest (ROI), your essential new source for global ‌financial commentary. Follow ROI on LinkedIn, and X.

And listen ‌to the Morning Bid daily podcast on Apple, Spotify, or the Reuters app. Subscribe to hear Reuters journalists discuss the biggest news in markets ​and finance seven days a week. (By Mike Dolan; Editing by Marguerita Choy)