Goldman Sachs just posted earnings per share of $20.98, roughly double what it made in the same quarter a year ago. Morgan Stanley’s profit jumped 58%. JPMorgan’s investment banking fees rose 30%, its equities revenue 86%. Across the five biggest US banks, combined quarterly profit landed close to $49 billion. Every headline out of this earnings season says some version of the same thing: Wall Street is back.
That’s true. It’s also incomplete. Strip away the growth percentages and look at what actually generated them, and a narrower story sits underneath the broad one. A story about how much of this quarter’s “recovery” runs through a small number of enormous, singular events rather than a wide base of ordinary dealmaking. That distinction matters more than the headline numbers do, because it changes what this boom is actually made of, and how it behaves if the conditions that produced it change.
The Numbers, All at Once
Worth laying out plainly before getting into what’s driving them. Goldman’s total net revenue hit $20.34 billion, up 39% year over year, with equities revenue alone at $7.42 billion, up 72%, and by the bank’s own account, the third straight quarter its equities desk had broken every prior record. Investment banking fees rose 55% to $3.40 billion, powered by a 130% jump in equity underwriting. JPMorgan’s numbers moved in the same direction: investment banking fees up 30%, equities revenue up 86%. Morgan Stanley reported record revenue of $21.35 billion, up 27%, with equities trading revenue hitting $6.3 billion about $1.9 billion above what analysts had modeled.
Citigroup’s quarter looked a little different, and the difference is worth noting before moving on. Net income rose 45% to $5.8 billion, and revenue hit $24.8 billion its best quarterly total in a decade — but the growth was spread across all five of its business lines rather than concentrated in one. Equities trading was actually the laggard: up 45% to $2.3 billion, which sounds strong until you realize it trailed the percentage gains at Goldman and JPMorgan. Citigroup’s own CFO acknowledged as much on the earnings call, saying the bank hadn’t built out its equities franchise fast enough to fully capture this quarter’s boom. That’s a useful contrast: Citigroup’s beat came from genuinely broad-based improvement across services, banking, wealth, and markets, while the banks posting the flashiest growth numbers got there largely by being more exposed to the handful of mega-deals driving the sector.
Zoom out to the industry level and the pattern holds at the aggregate: global investment banking revenue for the first half of 2026 came in around $61 billion, up roughly 24% from a year earlier. That’s the headline “Wall Street is back” number, and by itself it reads like a broad-based recovery across thousands of ordinary transactions the kind of steady rebound in corporate confidence that shows up gradually across an economy.
It isn’t quite that, for most of the banks posting the biggest headline gains.
One Deal, Five Winners
A meaningful share of this quarter’s outperformance traces back to a single transaction: the SpaceX IPO, which priced in June 2026 at roughly $85.7 billion, the largest public offering in history. Goldman, JPMorgan, and Morgan Stanley all held lead roles. The fees alone from that single deal ran to roughly $500 million split across the underwriting syndicate and the deal’s effects went well beyond the fee line. It generated a cascade of equities trading volume, prime brokerage income, and advisory work that shows up scattered across multiple revenue categories at every bank involved, making it hard to fully separate “SpaceX effect” from “genuine broad recovery” just by reading the income statement.
Then there’s what’s still coming. Goldman and Morgan Stanley are positioned to lead the upcoming Anthropic listing another AI-sector offering expected to draw the same kind of outsized attention SpaceX did. Citigroup picked up more than $70 million in fees as a joint global coordinator on the SK Hynix sale. And Bank of America alone has helped raise nearly $500 billion for AI-related companies since 2025, accounting for roughly 60% of all AI-linked fundraising across investment-grade debt, leveraged finance, and equity capital markets combined.
Put that together and a clearer picture forms: this isn’t diversified deal flow across a wide base of mid-sized companies feeling more confident about the economy. It’s a small number of extraordinary, correlated events nearly all of them tied to the same underlying theme of AI infrastructure buildout generating a disproportionate share of the sector’s reported growth. That’s a concentration problem hiding inside a recovery story. When five banks all beat estimates in the same quarter for what is substantially the same underlying reason, that’s not five independent signals confirming a trend. It’s closer to one signal, counted five times.
Why the Mix Itself Is the Story
There’s a second layer worth separating out from the deal concentration point: what kind of revenue is actually growing. For most of the post-2022 period, bank profits leaned heavily on net interest income the steady, unglamorous spread between what banks pay depositors and what they charge borrowers. That revenue is boring by design. It moves slowly, it’s driven by rate policy and loan volume, and it doesn’t spike or collapse in a single quarter.
What’s carrying results now is the opposite kind of revenue: fee income from trading, underwriting, and advisory work, all of which is inherently lumpy and deal-dependent. A single mega-IPO can move a bank’s entire quarterly equities line. A quiet quarter for large offerings can just as easily flatten it. Analysts covering the sector have pointed to exactly this shift Neville Javeri at Allspring Global Investments noted that capital markets and investment banking activity, not lending, have become the common driver behind every major bank’s beat this quarter.
That’s not a criticism of the banks; leaning into capital-markets fee income when the deal calendar is this rich is the rational move. But it does mean the sector’s current growth rate is riding on a more volatile, event-dependent revenue base than the growth rates of the past few years. The mix has shifted toward the kind of income that can disappear fastest.
Feast Now, What About Famine?
None of this is unfamiliar territory for investment banking. The sector has a long, well-documented history of feast-or-famine cycles the dot-com collapse gutted underwriting revenue for years, the 2008 financial crisis did far worse, and as recently as 2022–2023, rising rates froze the IPO market almost completely, leaving bank capital-markets desks with some of their leanest years in over a decade. Boom quarters like this one have shown up before. So has the reversal that tends to follow them.
The genuinely open question is whether this cycle is structurally different, because of what’s driving it. Prior booms were largely sentiment- and rate-driven cheap money and investor optimism pulling companies toward the public markets, then pulling back sharply when either one soured. This one is riding a multi-year AI infrastructure capital expenditure supercycle, which is a slower-moving, more durable kind of demand if AI capex spending itself holds up over the next several years. That’s the optimistic read: this isn’t investors chasing sentiment, it’s financing for an infrastructure buildout with years of runway left.
The more cautious read is that durability at the level of “AI capex broadly” doesn’t guarantee durability at the level of “bank fee income specifically.” Because so much of this quarter’s growth is concentrated in a handful of singular events rather than distributed dealmaking, the sector’s near-term results are unusually sensitive to the fate of a small number of specific transactions. If the Anthropic listing underperforms expectations, or the pace of AI-linked debt and equity issuance simply slows for a quarter or two, the earnings comparisons get a lot less flattering — not because the broader economy weakened, but because the mega-deal supply that’s been carrying results thinned out. A concentrated boom has a shorter list of things that need to go wrong to end it than a broad-based one does.
Who’s Better Positioned If the Deal Calendar Thins Out
Not every bank is exposed to that risk equally, and the difference comes down to business mix. Goldman’s model is the most capital-markets-heavy of the group, which is exactly why its equities growth outpaced peers this quarter and why it would likely see the sharpest pullback if mega-deal flow slows. Morgan Stanley’s business leans more heavily on wealth management alongside its investment bank, giving it a steadier, fee-based cushion that isn’t as tied to any single quarter’s deal calendar. JPMorgan’s universal-bank structure spanning consumer banking, commercial lending, and markets gives it the broadest diversification of the group, which is part of why its investment banking fee growth, while strong, wasn’t quite as extreme as Goldman’s equities number.
Citigroup sits at the other end of that spectrum. Its multi-year turnaround has deliberately spread growth across services, wealth, banking, and markets rather than betting heavily on any single business line which is exactly why its quarter looked broad-based instead of deal-driven, and also why its equities desk, still playing catch-up after years of underinvestment, didn’t capture as much of the SpaceX-and-AI-financing wave as its rivals did. In a quarter where being under-exposed to mega-deals actually helped smooth out the results, Citigroup’s slower, steadier build looks less like a laggard story and more like a different risk profile entirely.
None of that makes Goldman’s quarter less real. It makes it more concentrated a higher-beta bet on the same theme that’s currently driving nearly everyone’s numbers.
The Ground Floor Take
The instinct with a quarter like this is to read it as confirmation that the broader economy corporate confidence, deal appetite, capital markets health has genuinely turned a corner. Some of that is true. But a closer read of where the growth actually came from tells a narrower story: a handful of enormous, AI-linked transactions did a disproportionate amount of the work, and five banks reporting similar-looking beats in the same quarter isn’t five separate confirmations of a trend so much as five different exposures to the same underlying event calendar.
That’s not a reason to dismiss the results as the money is real, and the fees were earned. It’s a reason to watch what happens to bank earnings the first quarter the AI-linked deal pipeline goes quiet, rather than assuming this growth rate is now the new baseline. The SpaceX IPO alone reshaped a quarter of Wall Street earnings; a sector that can be moved that dramatically by a handful of listings is one where “record quarter” and “structurally different business” aren’t automatically the same claim, however much this earnings season made them sound like it. Whether that distinction matters much longer may depend on the same question we’ve been asking about the AI investment cycle all along: how much of this is durable infrastructure spending, and how much of it is one very large, very connected boom looking for its next mega-deal to keep the story going.
