Cyrus Reed: Iris, hey — tell me you read the Goldman earnings release before nine this morning, tell me I'm not the only one who did that.
Iris Holm: I read it at six-forty-seven. Specifically because I wanted to get there before the takes did.
Cyrus Reed: Okay, and what was your first — like, what hit you first, because mine was the headline revenue number, $20.34 billion, highest in Goldman's 157-year history, and I thought 'well that's the story,' but then I kept reading and wait — something else is going on here.
Iris Holm: The EPS beat. That's what hit me. Goldman's diluted EPS came in at $20.98. Analyst consensus was $14.48 to $14.50. A 45% overshoot. And it wasn't an isolated miss.
Cyrus Reed: Hold on — JPMorgan and Bank of America also missed consensus by wide margins the same quarter?
Iris Holm: Three institutions. Three independent models. Same quarter. All wrong in the same direction by a lot. Now, Goldman's equities desk printed $7.42 billion — all-time record for any bank division — and SpaceX priced the largest IPO in history with Goldman as lead underwriter. Those facts connect.
Cyrus Reed: So SpaceX is the — what, the detonator that broke the consensus models?
Iris Holm: That's where I want to land today. Goldman executed. But the story that nobody's telling is that three banks missing by this much simultaneously isn't a bank story at all — it's a forecasting infrastructure failure, live, in public.
Cyrus Reed: But wait — 'detonator' isn't quite right, because a detonator just sets something off. What SpaceX actually did is more like... okay, imagine a single concert sells out a stadium. Not just the venue makes money — every parking lot, every bar, every cab company in the city has its best night ever, simultaneously. That's what one deal did to Wall Street last quarter.
Iris Holm: The mechanism.
Cyrus Reed: Right — so the SpaceX IPO didn't just generate underwriting fees. It generated soft-dollar trading commissions from hedge funds who were basically paying banks in trading business to get IPO allocations, AND it triggered secondary market activity all at once. Three revenue lines. One event. Denis Coleman — Goldman's CFO — literally named SpaceX and Alphabet's equity raise on the earnings call as the primary drivers.
Iris Holm: Equity underwriting revenues up 130% year over year. $985 million. That's the line that tells you the mix shifted, not just the total.
Cyrus Reed: And it wasn't only Goldman catching it — JPMorgan earned $75 million just as a lead bookrunner on SpaceX, their investment banking fees hit $3.3 billion, up 30%. Bank of America, Morgan Stanley, both pulled downstream commissions from shared underwriting roles. The same concert filled every bar in the whole city, not just one.
Iris Holm: So the models failed because they're built for normal nights.
Cyrus Reed: Exactly — wait, no, that's actually the point I want to push on. Because KBW's Chris McGratty projected investment banking revenue up 26% and trading up 14% before earnings dropped. That's not a bad forecast — that's a structurally different event hitting every lever at once. No model assumes one deal cascades across underwriting, soft-dollar commissions, and secondary trading simultaneously.
Iris Holm: Fair. The forecasting failure isn't incompetence. It's architecture — models don't have a variable for 'gravity well deal.' Goldman's investment banking fees hit $3.40 billion, up 55% year over year. That's not a percentage any model was holding.
Cyrus Reed: And the scary part is — the conditions that made SpaceX that gravity well? Geopolitical volatility, allocation scarcity, concentrated capital demand — those don't just persist on command.
Iris Holm: And that's exactly where the model breaks. The equities beat alone — $2.3 billion above estimates — was larger than many banks' entire quarterly trading revenue. Not the total Goldman beat. Just the miss on one line.
Cyrus Reed: Wait — the beat on a single desk was bigger than what some banks print from their whole trading operation?
Iris Holm: That's the number. And total Goldman revenue came in $4.1 billion above forecasts. So McGratty's 14% trading growth call — it wasn't wrong because he miscounted. It was wrong because the variable he needed doesn't exist in a standard model.
Cyrus Reed: Okay but here's the — wait, actually the number is this: Goldman's equities desk, one desk, $7.42 billion. JPMorgan's entire investment banking division, all of it, $3.3 billion. One desk at one bank outearned another bank's whole advisory operation by more than two times. That's not a beat. That's a different sport.
Iris Holm: It's the specific fact that makes the hot take stick. Because if one line at one firm can do that, the consensus infrastructure has no ceiling variable.
Cyrus Reed: And net income hit $6.63 billion, up 78% year over year — David Solomon is citing AI infrastructure demand and accelerating strategic dealmaking, which, I mean, those are real things, but are they the *why* here or is he just — is he narrativizing a confluence?
Iris Holm: Both. The conditions were real. But what he's not saying — and what the CEOs actually said versus what they signaled — that's the part we need to get to, because Dimon's 'not repeatable' caveat sitting next to a record profit is a specific tension worth pulling apart.
Cyrus Reed: Right — but the part that doesn't fit is Solomon naming AI demand as a structural driver when the number was built on a single gravity-well IPO. Those are different theses.
Iris Holm: The kernel of the hot take survives. Three institutions, models built for a normal distribution, a $4.1 billion total miss. That's not noise.
Cyrus Reed: But that's what makes the CEO language so weird — because Dimon comes out and says conditions are 'close to as good as it gets,' which is like, that's a superlative, that's a peak claim, and then in the same breath he's basically saying don't build a model on this, large deals contributed meaningfully and results may not be repeatable. Those two sentences can't both be reassuring.
Iris Holm: They're not meant to be. He's protecting against two different audiences simultaneously.
Cyrus Reed: Okay — and then Jane Fraser comes in anticipating a 'summer lull' and flagging geopolitics as a wild card for the deal pipeline. Which, wait, that's not hedging, that's actually a forecast. She's saying the window closes.
Iris Holm: And Brian Moynihan names the Iran war specifically. Celebrates the quarter, then flags Iran as an active threat to the IPO pipeline. That's the tell.
Cyrus Reed: Wait — because the Iran war is what drove the volatility that made trading revenue spike. So it's the same variable pointing in two directions at once.
Iris Holm: That's the exact asymmetry. Geopolitical volatility inflated Goldman's $7.42 billion equities number — traders love uncertainty. But that same uncertainty is what kills CEO confidence in signing a $10 billion deal into the pipeline. Moynihan is celebrating the symptom and warning about the cause.
Cyrus Reed: So the defensible claim is — actually, no, let me try to say this right — the record is structurally real, meaning Goldman's desk actually printed $7.42 billion, that happened, but the floor that number implies? That's not real. Three tailwinds stacked: the SpaceX IPO window, geopolitical volatility boosting trading, AI capex demand pulling deals forward. None of them are stable. And the CEOs are telling you that directly while simultaneously handing you the trophy.
Iris Holm: That's the calibrated version. The record is real. The floor is not. Fraser, Moynihan, Dimon — three different banks, same signal. They know exactly what built this quarter. They're just not leading with it.
Cyrus Reed: And in 157 years — I mean, that's the number I can't shake. In Goldman's entire existence, no quarter looked like this one. Not 2008, not the post-financial-crisis bounce, not any of it. Q2 2026 is just... sitting there alone.
Iris Holm: Fine. Maybe it wasn't purely a forecasting failure. Maybe Goldman really is just that good at being in the right place when history decides to happen. I'll half-concede that. But the question that actually matters now — the one nobody running that desk can answer — is whether Q2 2026 was the first frame of a new cycle or the last frame of a perfect storm. And right now, even David Solomon isn't saying which one.
Cyrus Reed: No, and that's — wait, that's actually the most honest thing about this whole quarter. The people who built the record don't know if it repeats. The models that missed it don't know how to price the next one. Everyone's holding a trophy and squinting at the horizon.
Iris Holm: That's where I'll leave it. Three more quarters tell us whether Q2 was a regime or a moment. Good conversation.
