The 'tech rally' is narrower than it looks — that's the risk this week

Tech firms rallied, the headline said. In Seoul the chip names ripped — Samsung up 5.7%, SK Hynix up 8.3%, SoftBank up more than 11% in Tokyo. A proper AI party. Yet the same morning, Hong Kong, Singapore, Mumbai and a string of other markets fell. One rally, many markets, no agreement.

The U.S. tape tells the same split in miniature. The Nasdaq-tracking QQQ is up a fraction this morning; the broad S&P 500 is down; and the equal-weight S&P 500 — the index that treats every company the same size — is down too. Over the past week the equal-weight index lost roughly three-quarters of a percent while the tech-heavy index gained. Two benchmarks, pointed in opposite directions. That gap is not background noise. It is the story of this week.

Why this week matters: inflation is back as the thing that moves markets. A forecast-busting U.S. jobs report on Friday hardened bets that the Federal Reserve will raise rates at its September 16 meeting. The August CPI, due this Friday, is the last big input before that decision — a hot core reading is expected to cement a hike. Fed Chair Kevin Warsh spent his Jackson Hole speech warning on inflation and recommitting to the 2% target. The market is braced for a possible increase, the mirror image of the long era when the only question was how fast the Fed would cut.

Now the mechanics. A rate hike is a discount-rate shock, and it lands hardest on the assets whose payoffs sit furthest in the future — which is exactly the long-duration tech complex. So a tightening Fed should, on the surface, hit tech the worst. Instead the leaders are holding up and the average stock is sagging. Why? Because the AI mega-caps aren't trading on the rate; they're trading on an earnings machine that is still visibly accelerating. Nvidia, the biggest of them, doubled its revenue and is now a $4 trillion company, up again this morning. A leader has a story big enough to ignore the macro. The average S&P stock does not.

That is what makes this a concentration mirage rather than a real rally. A cap-weighted index can climb while a clear majority of its members weaken, because the few giants at the top do all the lifting. The equal-weight index removes that illusion. When cap-weight rises and equal-weight falls, the market is telling you the gains are borrowed from a handful of names and the breadth underneath is negative.

Now look at how the market is positioned for exactly that risk. In the options market, investors are buying protection: on the S&P 500 ETF there are roughly two-and-a-half puts outstanding for every call. That is not the footprint of a market building a confident, broad advance into a Fed decision. That is a market hedged for a two-way event.

Grant the best case — the AI capex cycle is real, the revenue is actual, and these specific businesses can keep climbing even if the Fed does move. That is defensible. But the index's strength is borrowed from them. The moment the leader stumbles — a disappointing Oracle print Thursday, or a CPI hot enough to finally tip even the AI group through risk appetite — there's no participation underneath to catch the broad index. That's the difference between strength and fragility. When equal-weight is bleeding while cap-weight holds, the market isn't broad and strong; it's narrow and held up by a single thread.

The condition that would change the reading is equally specific. Friday's CPI comes in soft, and — more tellingly — the equal-weight index starts participating again, everyday stocks catching up to the leaders. That is dispersion resolving, and it would make the rally real. A hot CPI, with equal-weight still lagging while the tech leaders take a leg down, is the unwind of everything the headline currently sells. Watch the breadth, not the headline.