Bouts of volatility in the artificial intelligence trade are threatening to snowball into a rout as positioning and crowding levels flash red, with traders steeling themselves for a rollercoaster ride into the summer.
While the long-term AI story remains intact, the immediate picture has become considerably murky. Stretched positioning, massive leverage in exchange-traded funds (ETFs), options-related hedging activity and a sharp rise in both prices and swings across semiconductors are creating a treacherous landscape.
Hedge Funds Start to Trim AI Bets
Whether it's active or passive funds, hedge funds or quantitative strategies, institutions or retail investors, everyone is effectively getting longer AI by the day, according to Goldman Sachs Group Inc. partner Bobby Molavi. “The co-correlation of strategies is a point to watch,” he said. “Feels great on the way up, but can be extremely dangerous on the unwind.”
Prime brokers, including those at UBS Group AG, are flagging that some hedge funds have started reducing exposure to crowded trades in a bid to diversify. This shift comes as doubts about the future returns on vast investments by big tech hyperscalers return to the forefront.
Leveraged ETFs Amplify Risks
One of the biggest worries is the rapid proliferation of leveraged ETFs. Accounting for about US$200 billion in assets, every one per cent move in the underlying index triggers roughly US$9 billion of rebalancing in the same direction due to a significant short-gamma effect. Technology and momentum-driven strategies account for roughly 85 per cent of these ETFs, exacerbating swings across the sector.
For Nomura cross-asset strategist Charlie McElligott, the AI trade has become a perpetual motion machine. The unusual combination of rising prices and higher volatility across areas such as chips, memory, servers, opticals and networking is feeding market weightings, creating a powerful feedback loop. Leveraged ETFs are buying the upside, a combination that “exposes us to collapsing under the weight of the delta on even just profit-taking-turned-waterfall,” McElligott said.
Comparison to the Dot-Com Era
The current market is beginning to feel eerily similar to the final months of the dot-com era, Molavi said, when investors typically got used to taking sudden five per cent moves in their stride. This raises the question of “what happens if 10 per cent breaks, and whether there is no floor in sight after that,” he said.
Doubts about AI are likely to reinforce the need for rotation into the next trade. Realized volatility has been rising sharply as well. A continuation of sharp moves like those seen this week is likely to turn volatility-control funds into heavy sellers, according to McElligott. He estimates US$21 billion of selling flows on two weeks of daily one per cent moves in the S&P 500, and US$41 billion on 1.5 per cent moves.
