Jonathan Ingles: Cole, good to be back — I want to start with a scene, bear with me.
Jonathan Ingles: September 2026. Retired teacher, logs in, sees her S&P 500 index fund up. Feels like she did everything right — broad, diversified, the whole market. Same morning, Michael Cembalest at J.P. Morgan Asset Management publishes his Eye on the Market piece. Forty-two stocks. Sixty to eighty percent of all earnings growth, price returns, capital spending in that index since January 2024. She doesn't own the market. She owns forty-two companies and four hundred and fifty-eight bystanders.
Cole Bryant: That's — okay, I mean, that actually reframes every conversation I've ever had about index investing. What's the plain version of what went wrong?
Jonathan Ingles: Nothing went wrong yet. That's the uncomfortable part. The Kobeissi Letter ran the same forty-two names — sixty-seven percent of the S&P's thirty-three percent total gain since early 2024, twenty-two percentage points of index return, those forty-two stocks. It's a market-cap-weighted basket, and the heaviest items — Nvidia, Microsoft — are doing almost all the lifting. Think of it like a grocery basket where two products are eighty percent of the weight. You're not buying groceries. You're buying those two things.
Cole Bryant: No but wait — okay, so when was the last time we saw a basket this lopsided? Because I have a number and it's not 1999.
Cole Bryant: Index concentration at its highest since 1927 — that's a hundred-year record — and the teacher logging in has no idea.
Jonathan Ingles: And she still thinks the headline is real. That's where Ioannis Blekos at Goldman Sachs comes in — he flagged roughly $150 billion in paper gains sitting inside Q2 2026 S&P 500 EPS. Strip those out, year-over-year growth is still 26%. Which is strong. But it's not the headline figure. The number is real, just smaller than advertised.
Cole Bryant: Wait — $150 billion in paper gains just... baked into the earnings number?
Jonathan Ingles: Baked in. And here's what makes it worse — the money is flowing uphill. Nvidia, Microsoft, the infrastructure layer — they're capturing it. The companies actually deploying AI, burning through that $800 billion capex? Nothing yet in their earnings.
Cole Bryant: That's — wait, no, that's the picks-and-shovels thing playing out in real time. The gold rush is happening and the only people making money are selling shovels. The miners are just... spending. And I mean, Goldman's breadth measure is already at dot-com lows, right? Like 60% of S&P 500 stocks are more than 20% below their all-time highs while the index is sitting near records. The median stock is 16% below its 52-week high. Tech's share of U.S. GDP growth went from basically zero in 2022 to 33% — but none of that is reaching the companies actually adopting this stuff yet.
Jonathan Ingles: Only about 30% of stocks are even outperforming the index on a three-year basis. Lowest reading since 1991. The index is a mask.
Cole Bryant: And the mask stays on as long as Nvidia keeps printing. But what happens when someone asks the adopters to prove it?
Jonathan Ingles: That's exactly where Cembalest's qualified optimism and Goldman's structural warnings stop making sense together — and why the 2027-2028 earnings cycle is the real test nobody wants to talk about yet.
Cole Bryant: Okay but that's — wait, that's the thing that gets me about Cembalest's actual words. He says he doesn't expect the AI theme to fail in the next six to twelve months. That's a very specific window. Is that a prediction or is that just... liability coverage?
Jonathan Ingles: It's both. The caveat is doing enormous work. He's ringing the alarm in Eye on the Market and simultaneously telling the people paying J.P. Morgan Asset Management not to panic for two quarters. That's not a forecast. That's a managed timeline.
Cole Bryant: No way he'd phrase it that way.
Jonathan Ingles: He doesn't have to. The asymmetry is in what he left unsaid — a slowdown in AI produces, his words, significant repercussions across markets and the broader economy. That's not sectoral. That's systemic. And Goldman's warnings are already present in the data right now, not in twelve months. State Street Global Advisors pushes back — they argue effective S&P 500 diversification is still adequate, well above levels that historically signal real danger. Which, I mean — maybe. But Cembalest and Goldman aren't fully reconcilable with that view. Someone's methodology is flattering the picture.
Cole Bryant: Where does IBM fit — because you mentioned it earlier and I want to hear the actual comparison.
Jonathan Ingles: IBM in the sixties and seventies commanded roughly the same market concentration Nvidia holds now. The computer revolution was completely real. IBM's dominance wasn't an illusion. The stock still got halved — not because computers failed, but because the supplier edge got commoditized. The picks-and-shovels trade collapsed even though the gold rush continued. That's the parallel. Chelikavada and Bennett flagged it in a September 2025 arXiv paper — scientific publishing patterns before the dot-com era look structurally similar to now. Contested analogy, I'll grant that. But the IBM case alone should give Nvidia bulls a specific thing to sit with.
Cole Bryant: So the 2027-2028 earnings cycle is basically — that's when the hyperscalers either show the productivity multiplier or they don't. And Cembalest's six-to-twelve months runs out way before we even get there.
Jonathan Ingles: Exactly. And Cembalest's window closes well before that CFO sits down. That's the gap nobody's pricing. The 42-stock concentration either gets vindicated by a productivity line that actually moved, or it turns out the whole thing was a crowding trade into a narrowing set of suppliers. Not a revolution. A very expensive bet on a promise.
Cole Bryant: I mean — that teacher who logged in September 2026 and felt fine? She's still logging in. Same fund. Same number going up. And she still doesn't know if she owns a revolution or just... Nvidia's next quarter.
Jonathan Ingles: That's where we are. The answer comes in 2027. Not now.
Cole Bryant: Good place to sit with it. Thanks for walking through all of it.
