Goldman's Solomon Says AI Won't Wipe Out White Collar Jobs. I've Heard This Before.

Altimeter's Apoorv Agrawal called AI "one of the largest capital formation cycles" he's seen. Mary Daly from the San Francisco Fed also weighed in on AI's impact on productivity and labour markets at Bloomberg Tech this week, though the details of her comments weren't fully reported.

The capital flowing into AI companies is staggering. SpaceX and Anthropic going public will be landmark events. Solomon said it's "good for the US to have the biggest, most important companies in the world." Fair enough. Hard to argue with that as a general principle.

But I called my old colleague at Siemens last week, and he made a point that stuck with me. The companies building AI and the companies deploying AI are not the same. The returns might be very different too. Nvidia sells shovels. Goldman buys shovels. Different business.

Should we believe the "no wipeout" prediction?

I don't know. That's the honest answer.

Solomon has access to data I don't have. He's running a bank with tens of thousands of employees and watching productivity metrics in real time. If he says Goldman's headcount isn't collapsing, I believe him about Goldman.

But here's what remains unclear: what happens to the firms that aren't Goldman? The mid-tier accounting practices, the regional law firms, the back-office processing centres that don't have Goldman's margins or Goldman's talent? Those are the places where, in my experience, automation hits hardest. Not at the top. At the middle.

When we installed robot arms at automotive plants, the first jobs that disappeared weren't the skilled technicians. They were the material handlers, the quality inspectors doing repetitive checks, the workers doing tasks that were dangerous or boring or both. The skilled people often stayed. Sometimes they even got more interesting work. But the workforce shrank. That's just, well, that's just what happened.

What's actually useful here?

The most interesting thing Solomon mentioned was how AI is being deployed across different levels of the bank. Back-office employees, junior analysts, senior bankers, all using the same underlying technology but in different ways. That's a useful case study.

It suggests AI's impact won't be uniform. Some roles get augmented. Some get compressed. Some disappear entirely. The pattern probably depends on how much of the job is pattern-matching versus genuine judgment.

I don't have a strong prediction here. Anyone who tells you they know exactly how this plays out is selling something. But I'd bet the impact looks different at Goldman than it does at your local insurance broker. And I'd bet the timeline is longer than the optimists think and shorter than the skeptics hope.

Solomon's doing his job, talking up the economy and his bank's role in it. That's fine. But take the "no wipeout" framing with a grain of salt. The wipeout, if it comes, probably won't look like mass layoffs at Goldman Sachs. It'll look like smaller firms never hiring the people they would have hired. Quieter than a wipeout. But real.

James Chen · 10 hours ago · 4 min