Goldman sees five-year deal backlog, says AI won’t cut headcount

Goldman Sachs said its deal backlog was at a five-year high, a sign David Solomon used to argue that the bank’s near-term revenue pipeline remains full even as artificial intelligence spreads deeper into daily work. Solomon said clients are turning to Goldman to lead their most strategic and consequential transactions, and said he expects the firm’s “flywheel of activity” to keep turning as those pipelines convert into fees.

The comments came as Goldman reported second-quarter 2026 net revenues of $20.34 billion and net earnings of $6.63 billion, with diluted earnings per share of $20.98 and annualized return on equity of 23.5%. Goldman’s stock rose 9% on the day of the earnings coverage, underscoring how strongly investors responded to the mix of a packed deal pipeline and firm earnings power.

For bankers inside Goldman, the backlog matters because it usually means longer queues of mandates, more live pitches, and more pressure on execution teams to move faster without dropping the ball. In practice, that can mean analysts spending more time cleaning models and drafting materials, associates absorbing more last-minute revisions, and vice presidents trying to keep deals moving while preserving enough judgment to catch problems before they surface. If Solomon is right that AI should make people more productive but should not affect headcount, the immediate effect is likely to be workflow compression rather than fewer seats.

Goldman’s own AI messaging points in that direction. In February, the firm said it was working with Anthropic on AI agents for trade accounting and client onboarding, with Marco Argenti framing the effort as an efficiency play rather than a near-term job cut story. The bank described the systems as a kind of “digital co-worker” for process-heavy roles, a sign that automation is already being aimed at the work that clogs back offices and slows front-office follow-through.

That sits inside a broader corporate strategy Goldman has already labeled “One Goldman Sachs 3.0,” its AI-driven operating model introduced in the 2025 annual report. The same report said 2025 net revenues reached $58.3 billion, up 9% from a year earlier. Goldman Sachs Research has also warned that AI could displace 6% to 7% of the U.S. workforce in a broad adoption scenario, while estimating that fully adopted generative AI could lift U.S. labor productivity by around 15%. For Goldman staff, the tension is no longer whether AI arrives, but whether the firm uses a five-year backlog to stretch teams further, or to give them more room to grow before the next promotion cycle.