JPMorgan Chase CEO Jamie Dimon Thinks AI Spending Is Going to Reach $1 Trillion Next Year

The market’s real takeaway is not that the big-bank earnings were strong; it’s that fee-sensitive franchises are now the cleanest public-market proxy for whether the AI/risk-on cycle is still accelerating. That favors Goldman over more diversified lenders because underwriting and trading leverage turns a modest increase in issuance into outsized EPS revisions, while buybacks give downside support if the cycle cools. The second-order winner is the exchange/listing complex: if capital formation stays open, NDAQ and adjacent market infrastructure names benefit from the broader reopening of IPOs, follow-ons, and secondary volume.

The AI spend message matters more for semis and data-center infrastructure than for banks themselves. If corporate capex is truly moving toward $1T+, the incremental earnings pool accrues to NVDA first and then to the network/power stack around it; banks only capture the financing and advisory tolls. That creates a subtle but important spread trade: capital-markets exposure is a near-term beta play, while AI hardware demand is the longer-duration monetization path.

Contrarian risk: consensus is extrapolating a straight-line capex curve, but these cycles usually front-load and then digest once procurement is complete or financing gets tighter. The thesis breaks if IPO volume decelerates, if deal pipelines fail to convert into closed fees over the next 1-2 quarters, or if hyperscaler capex commentary rolls over into the next earnings season. In that case, GS’s operating leverage works in reverse faster than JPM’s more diversified earnings mix, and NVDA’s multiple becomes more vulnerable than its fundamentals.