“I’m not gonna try to play semi- / AI- expert here,” Nomura’s Charlie McElligott said Monday, amid a burgeoning panic across the tradable AI space, where the leadership struggled under the weight of flashy headlines touting the capabilities of “DeepSeek,” a low-cost Chinese competitor whose shoestring budget raised uncomfortable questions about indulgent hyperscaler capex.

McElligott may not be an expert on AI, but he’s an expert on flows, and as he wrote Monday, a sudden drawdown in the Mag7 leadership has the potential to waterfall, snowball and otherwise escalate courtesy of modern market structure.

“The secular escalatory shock here is the potential that DeepSeek’s ‘autonomous reasoning’ model is taking a large leap in pushing us closer towards recursive self-improving AI [without] requiring the massive semiconductor spend of the newest and most expensive chips,” Charlie wrote. That, in turn, “threaten[s] prior assumptions [about] the AI ecosystem, with massive implications [for] pricing power, capex trends and market valuations of the current hierarchy of leadership in the space,” he went on.

Speaking in generalities, the current US-centric macro-market conjuncture’s a giant perpetual motion machine where the narrative drives inflows and the inflows swell market caps and swollen market caps beget more inflows and the resultant wealth creation drives the consumer spending which propels the US economy. And around and around. If something throws a spanner in the works, it’s not just a few names at risk, but the whole interlocking, self-perpetuating system.

“AI / mega-cap tech / Mag7-8 has become ‘THE’ US stock market from an index return attribution perspective, which forced anybody with a global equities mandate to chase and try to over-own these exposures if you were going to try to beat — or simply keep pace with — benchmarks, so this is where we need to watch the risk of actual long-term reallocation flows from foreign investors in particular,” McElligott said.

Remember: There really isn’t a “global” equity market anymore. There’s just the US equity market, and as Charlie alluded to, within US equities, there are only the top seven to 10 names. As discussed here earlier this month, US mega-cap dominance has reached what I called “existential extremes,” and for once, “existential” isn’t hyperbole.

The charts above (click to enlarge, as always) give you a sense of just how extreme the flow dynamic really is in US tech and large-caps.

When a market’s this concentrated — when breadth’s this poor and the leadership’s this narrow — the margin for error is vanishingly thin. The US now comprises nearly 75% of the MSCI World Developed Market index, and the top names are 35% of US market cap.

“The lack of breadth in stocks from such a prolonged period of leadership via a small group of companies can pose an outright stability risk on multiple fronts,” McElligott wrote, on the way to enumerating, as follows:

1. Huge mega-cap tech weightings = problems at the equity index level thanks to crowded single-name ownership among institutions and retail;
2. Potential issues for dispersion / correlation traders; and
3. Down the road, knock-on [effects for] ‘vol laundering’ private investments in the AI space

There’s more. Regular readers might recall that since a favorable US CPI report stanched the bleeding in bonds earlier this month, market participants began to evidence renewed “fear” of the right-tail, which is to say concern they might miss a new melt-up — “FOMO,” if you prefer to speak in acronyms. And where does FOMO show up? In the call wing. Call skew was bid post-CPI. That’d be relevant in a tech rout. “The market’s long a ton of delta here, which could then act as a negative delta de-risking flow as calls move out of the money and puts are chased,” Charlie said.

Oh, and don’t forget that a huge share of exploding leveraged equity ETF AUM is tied to tech, AI and semis. Have a look:

Out of some $125 billion of leveraged ETF AUM, nearly nine of every 10 dollars is, in some way, shape or form, related to AI.

Why does that matter? Well, it’s the implicit rebalancing needs. As McElligott explained, a big drawdown for the biggest tech names “would pose a massive ‘sell-to-rebalance’ flow risk” out of the leveraged ETF product space.

Panning back out to a 30,000-foot view, the entire macro-market zeitgeist in 2025 centers around US exceptionalism, and a very big part of that story is the assumption that America’s pole position in the AI race is unassailable. You know what they say about assuming.