Key Takeaways
- • Nomura's Charlie McElligott, in his cross-asset note "Cleansing," argues the AI complex is showing meaningful stabilisation after two weeks of punishing deleveraging, with Nasdaq options implied volatility declining across strikes and expiries over the preceding three sessions.
- • McElligott's long AI enablers and bottlenecks versus short hyperscalers strategy lost approximately 11% over the preceding quarter and carried a year-to-date drawdown of roughly 25.7%, but recovered about 6% over the latest three sessions.
- • Anthropic's IPO prospectus shows revenue rising from approximately $386 million in 2024 to $4.6 billion in 2025 and $16.3 billion in the first half of 2026, with revenue per dollar of compute spending improving from about $0.57 to $2.08 a year later.
- • Hedge fund Magnificent Seven exposure sits near the 96th percentile of its historical distribution while mutual fund exposure sits near the 10th percentile, a divergence McElligott sees as capable of amplifying another upside chase.
- • Momentum and carry strategies remain under strain from crowded positioning, and McElligott views recent hedge fund stop-outs as a necessary reduction of excess leverage rather than proof the investment model has failed.
Summary
Nomura's Charlie McElligott, in his latest cross-asset note titled "Cleansing," argues that recent sessions have delivered an important stabilisation in the AI complex following a sharp repricing of growth expectations and AI financing risk. McElligott describes October as a positioning cleanout in which investor confidence is returning faster than leverage is being repaired. He cautions that the easing of pressure does not mean the market is out of danger, since multi-strategy hedge funds continue to work through a destructive period of momentum reversals, crowded relative-value positioning and mechanical risk reduction.
The first signal of changing sentiment, according to McElligott's analysis, came from the options market. Nasdaq volatility has retreated across much of the options surface, with Nomura's QQQ volatility monitor showing implied volatility declining across a broad range of strikes and expiries over the preceding three sessions. Lower implied volatility reduces the cost of downside protection and can ease dealer hedging flows that amplify price moves.
McElligott points to his long AI enablers and bottlenecks versus short hyperscalers strategy as the clearest evidence of stabilisation. The trade lost approximately 11% over the preceding quarter and had a year-to-date drawdown of roughly 25.7%, then recovered about 6% over the latest three sessions, extending its weekly gain toward 8%. The trade reflects the distinction between companies supplying scarce compute, memory and infrastructure and the hyperscalers financing that expansion under margin and free cash flow pressure.
Three developments support that distinction, per McElligott. First, Anthropic's IPO prospectus shows revenue rising from approximately $386 million in 2024 to $4.6 billion in 2025 and $16.3 billion in the first half of 2026. McElligott highlights that revenue per dollar of compute and infrastructure spending improved from about $0.57 in the second quarter of 2025 to $2.08 a year later, and that the company shifted into quarterly operating profitability. He cautions that these results should not be extrapolated across the industry, given enormous capital requirements and intensifying competition. Second, investors absorbed Meta's initial equity offering alongside continuing large infrastructure financings, which McElligott cites as evidence that capital markets remain open to AI expansion. Third, SK Hynix reported revenue growth of 328% year on year, operating profit growth of 632% and operating margins near 80%, with production capacity reportedly committed into the latter part of the decade.
Positioning data also shows a sharp divergence. McElligott, drawing on Nomura's options flow observations, identified substantial upside call buying in Apple, Alphabet and Microsoft ahead of their earnings, including roughly 70,000 January Apple 380 calls, 50,000 January Alphabet 360 calls and 42,000 January Microsoft 585 calls, representing about $1.6 billion of estimated delta exposure. Nomura's beta monitor shows long/short equity hedge funds carrying Magnificent Seven exposure near the 96th percentile of their historical distribution, while mutual funds sit around the 10th percentile. Microsoft, Tesla, Nvidia and Amazon show particularly elevated hedge fund sensitivity, while Apple is a notable exception. McElligott's view is that this imbalance means strong megacap earnings could prompt underexposed mutual funds to rebuild positions.
The first signal of changing sentiment, according to McElligott's analysis, came from the options market. Nasdaq volatility has retreated across much of the options surface, with Nomura's QQQ volatility monitor showing implied volatility declining across a broad range of strikes and expiries over the preceding three sessions. Lower implied volatility reduces the cost of downside protection and can ease dealer hedging flows that amplify price moves.
McElligott points to his long AI enablers and bottlenecks versus short hyperscalers strategy as the clearest evidence of stabilisation. The trade lost approximately 11% over the preceding quarter and had a year-to-date drawdown of roughly 25.7%, then recovered about 6% over the latest three sessions, extending its weekly gain toward 8%. The trade reflects the distinction between companies supplying scarce compute, memory and infrastructure and the hyperscalers financing that expansion under margin and free cash flow pressure.
Three developments support that distinction, per McElligott. First, Anthropic's IPO prospectus shows revenue rising from approximately $386 million in 2024 to $4.6 billion in 2025 and $16.3 billion in the first half of 2026. McElligott highlights that revenue per dollar of compute and infrastructure spending improved from about $0.57 in the second quarter of 2025 to $2.08 a year later, and that the company shifted into quarterly operating profitability. He cautions that these results should not be extrapolated across the industry, given enormous capital requirements and intensifying competition. Second, investors absorbed Meta's initial equity offering alongside continuing large infrastructure financings, which McElligott cites as evidence that capital markets remain open to AI expansion. Third, SK Hynix reported revenue growth of 328% year on year, operating profit growth of 632% and operating margins near 80%, with production capacity reportedly committed into the latter part of the decade.
Positioning data also shows a sharp divergence. McElligott, drawing on Nomura's options flow observations, identified substantial upside call buying in Apple, Alphabet and Microsoft ahead of their earnings, including roughly 70,000 January Apple 380 calls, 50,000 January Alphabet 360 calls and 42,000 January Microsoft 585 calls, representing about $1.6 billion of estimated delta exposure. Nomura's beta monitor shows long/short equity hedge funds carrying Magnificent Seven exposure near the 96th percentile of their historical distribution, while mutual funds sit around the 10th percentile. Microsoft, Tesla, Nvidia and Amazon show particularly elevated hedge fund sensitivity, while Apple is a notable exception. McElligott's view is that this imbalance means strong megacap earnings could prompt underexposed mutual funds to rebuild positions.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.