Why are stocks struggling, where that just means struggling to make new(er) highs?

Don’t answer that, it’s a trick question.

Equity benchmarks are struggling because — wait for it — equity benchmarks are struggling.

The tautological nature of that assessment’s a function of extreme concentration. If the heaviest weights are in a correction, and some of them are, the cap-weighted benchmarks they dominate are commensurately hamstrung.

I alluded to the underlying problem on multiple occasions this week. On Thursday, in “Obvious Things,” I noted that looking back on it, we should’ve realized last year that hyper-scaler longs were passé, that semis were the next big thing (because “All that capex has to go somewhere” after all) and that Nvidia’s bleeding edge GPUs were just part of the semi story (AI systems are anatomical and bodies need more than a brain).

Of course, everything’s obvious after the fact. The Medallion Fund’s actually second on the list of all-time best-performing funds. Hindsight Capital’s number one, and always will be.

Long story short (no market pun intended), we should’ve sold the hyper-scalers and used the money to buy the semis. That’s what’s going on now, in 2026.

The result: Meaningful drag from the hyper-scalers which, as Nomura’s Charlie McElligott pointed out, are a source of funds for investors chasing the semi train and/or rotating into cyclicals.

“The US stock market’s currently chopping itself into pieces and unable to make new all-time highs without the participation of the Mag7 cohort, and in this case more specifically, the hyper-scalers,” Charlie wrote Friday.

The figure above shows you the “hyper-drag,” so to speak.

Those four names are simply too important from a figurative and literal “weight of the world” perspective vis-à-vis the indexes for the “broad” market to break out to even higher highs, Micron’s best efforts to play Atlas in 2026 notwithstanding.

It doesn’t help that hyper-scaler equity demand’s now hyper-scaler equity supply, which is to say netting buybacks against issuance gets you a negative number for Amazon, Google, Meta, Microsoft and Oracle in 2026.

As a group, the Magnificent 7’s down about 14% from the mid-May highs.

The simple figure above gives you some context for the current downtrade by including both the Iran war lows and the “Liberation Day” selloff in the lookback.

We’re hardly in trainwreck territory, but the overarching point is that it’ll be difficult to get a summer melt-up with the Mag7 dragging somewhere between a technical correction and bear market.

You can think of big AI spender “deficits” as semi heavyweight “surpluses,” but from a broad market performance perspective, that’s just left pocket to right pocket accounting. Unless and until the hyper-scalers rediscover their joie de vivre, cap-weighted benchmarks are going to struggle.

If the Mag7 falls much further, it could be a catalyst “for [a] proper risk-off summer,” BofA’s Michael Hartnett said, in his latest. For now, record-high margins remain a pillar for stocks, and “liquidity pulled from mega-cap AI arms racers is simply racing into semis and illiquid cyclicals to front-run a Trump pivot to affordability,” he wrote.

In the same Friday note mentioned above, McElligott noted that the hyper-scalers haven’t made a new high versus the equal-weighted S&P since October.

Investors, he said, are beginning to “get their arms around the idea that the AI trade’s moving from big spenders as the ‘long AI’ proxies to now instead a place where that capex equals precipitous cash burn, accelerated debt financing and increasing likelihood of major equity issuance.”

In the case of the latter, Charlie went on, the hyper-scalers would be “de facto overwriters, selling calls on their own stocks.”