Goldman Sachs and Morgan Stanley earnings provide a clear picture of Wall Street collecting its cut of virtually every relevant AI transaction
Charly Triballeau / AFP via Getty Images
Talk about a real-time snapshot: Bank earnings season, unfolding this week, has offered perhaps the clearest picture yet of which sectors of the economy are benefitting from this historic capex binge brought on by the AI buildout. Banking is, at it happens, not least among them. Goldman Sachs $GS and Morgan Stanley $MS results together provide a particularly clear picture of Wall Street collecting its cut of virtually every relevant transaction.
Unsurprisingly, the primary financing mechanism is debt. Major tech companies have announced more than $700 billion in capital expenditures for 2026 alone — an eye-watering 70% increase even over last year’s historic numbers — and with such enormous spending, there’s simply no way that cash could cover it all, were it even the case that corporations wanted to finance the buildout that way. This means bond markets are a major channel, and growing at similarly impressive rates, year over year.
As an example, the five major AI hyperscalers issued $121 billion in U.S. corporate bonds in 2025, compared to an average of $28 billion per year in the five years prior. Estimates for 2026 range now from $175 billion to $300 billion, which means the question is whether the market will grow a mere 50% or, at the upper end of the range, nearly triple.
Of course, someone has to help out by underwriting those bonds, trading them, advising on the deals, and managing the resulting assets (not to mention helping newly wealthy AI founders manage their newfound wealth). Enter Wall Street's biggest firms. Here's an even more granular look at the dynamics.
Goldman's first-quarter results function like receipts for the overall trend. Investment banking fees came in at $2.84 billion, up nearly 50% from just a year ago. Advisory alone hit $1.5 billion, up nearly 100%, driven by AI-related M&A. Equity underwriting contributed $535 million, up 45%, much of it from convertible bonds — a structure AI companies favor because the equity upside allows them to borrow at artificially cheap rates. Debt underwriting added $811 million.
The Oracle $ORCL deal in February in particular shows how a single transaction generates revenue at multiple points along the chain. The database giant — notably the weakest credit bet among the AI hyperscalers, its bonds already trading closer to junk than investment grade — needed $25 billion. Goldman helped run the deal. Investors placed $129 billion in orders, the largest order book ever recorded for a corporate bond offering of that size.
Describing the deal to Bloomberg afterward, Goldman's head of investment-grade syndicate in the Americas, John Sales, said: "Everybody's talking about record supply. I think the more interesting story is record demand."
Goldman also reported record equities revenue of $5.33 billion, up 27%, driven by prime brokerage — lending to the hedge funds and institutional investors who are also, by and large, making AI-related trades. Total assets under supervision reached $3.65 trillion.
Goldman shares, for their own part, are up 80% over the past year, broadly in line with the business’s sharp upward trend.
Morgan Stanley reported record net revenues of nearly $21 billion for the quarter, up 16% year over year, with institutional securities — its investment banking and trading arm — delivering record revenues of roughly $11 billion. Fixed income underwriting was up 36%, with the release specifically citing "higher investment grade issuances benefitting from increased event-related activity." That’s the AI debt wave in a single earnings phrase.
Morgan Stanley shares are up 75% over the past year, nearly on par with Goldman’s for what amounts to a handy 12-month double.
What makes Morgan Stanley's position most interesting and emblematic is the circularity, research driving results and vice versa. Last month, the firm published a memo arguing that AI is "poised to be a net positive for the banking industry," that banks stand to benefit directly from the AI capex boom, and that AI could boost bank productivity by 20 to 50% over the next five to 10 years.
Now, with its quarterly results, Morgan Stanley is simultaneously describing in footnotes how it’s cut approximately 2% of its global workforce in a March "workforce management action" — severance costs of $178 million — which it frames as an efficiency initiative. The firm is collecting fees on the AI buildout, publishing research explaining why that's good for banks, and simultaneously running an AI efficiency experiment on its own business. All in the same quarter, too.
The final piece of the picture is how these moves pay off for the banks’ shareholders, its own management and executives among them, with Morgan Stanley stock outperforming the S&P 500 over the last year by 45 percentage points.
The KBW Nasdaq $NDAQ Bank Index — an imperfect proxy, given it includes regional and consumer players much less exposed to capital-markets activity — is nonetheless up 54% over the past year, a figure that still reflects how AI’s broad benefits play out across the sector.
JPMorgan $JPM's Jamie Dimon was direct in his quarterly remarks, listing "AI-driven capital investment" alongside fiscal stimulus and deregulation as forces playing well for the bulge bracket. That's text and subtext at once, demonstrating how the AI buildout has become so large, and so dependent on debt, that it is now among the most meaningful macroeconomic variables, with the banks that facilitate it among its primary beneficiaries.
Join 500,000+ readers who start their day with Quartz.
By subscribing, you agree to our Terms of Service and Privacy Policy.
