Ken Griffin Went Home Fairly Depressed
In January, Ken Griffin called AI “garbage.” In May, he came home from work fairly depressed because AI agents were doing in hours what his PhD finance team does in months.
Four months. That’s the gap between those two statements. Same guy. Same firm. Same data feed into his head every morning.
Who said this
Ken Griffin is worth somewhere around $49.6 billion (Forbes) to $51.2 billion (Bloomberg) as of early 2026. He’s the 34th-richest person on the planet. He founded Citadel in 1990. Citadel LLC runs about $67 billion in AUM after returning $5 billion in profit to investors. He owns roughly 80% of both Citadel LLC and Citadel Securities, the market-making arm that handles a meaningful chunk of all US equity volume on any given day. His stake in Citadel Securities alone is worth around $17 billion.
He runs the firm from Miami. He moved it there from Chicago in 2022 because he didn’t like the tax math.
Griffin is one of the sharpest pure operators in finance. He runs a multi-strategy hedge fund and the largest options market-maker in the US, both at the same time. He pays for the best information money can buy and employs a small army of master’s and PhD-level quants to ground-truth every claim that hits his desk. He has no equity in any AI lab, no book to sell, no podcast to promote.
When a guy like that calls something garbage, that’s the position of the house. When he comes home depressed four months later, that’s also the position of the house.
What he said in January
At Davos, January 22, 2026, Griffin gave his read on the state of AI in three words: “it’s all garbage.”
That’s a flat dismissal. On a stage where every word he says gets quoted into a hundred research notes by Monday. With his name on it. He had room to hedge with “the hype is ahead of the substance” and chose not to use it.
That was the bear case from maybe the most credible bear on Wall Street.
What he said in May
Stanford GSB Leadership Forum, four months later. Same guy. Different read.
“In the last few months, there has been a step change function in the productivity of the AI toolkit. It is profoundly more powerful than it was just 9 months ago.”
The punch:
“Work that we would usually do with people with master’s and PhDs in finance over the course of weeks or months, being done by AI agents over the course of hours or days.”
Read that twice. The work product of Citadel’s most expensive headcount, the people who get hired out of MIT and Princeton and Wharton PhD programs, compressed from months to hours.
And then this:
“I went home one Friday, actually fairly depressed by this because you could just see how this was going to have such a dramatic impact on society.”
Fairly depressed. From the guy with the $17 billion market-making book and the staff that verifies every model against live order flow. That word choice matters. Most operators in his position default to “excited” or “bullish” or some version of “biggest opportunity of our lifetime.” He picked depressed.
And:
“When you’re seeing really high-level research being done by AI engines, it’s quite eye-opening.”
“High-level research” in Griffin’s vocabulary doesn’t mean a chatbot writing a memo. It means the kind of work he pays $300K starting comp to staff up.
The reversal is the story
Most “AI is here” content this year has come from people whose business model depends on AI being here. Tech founders, VCs, the consultants riding the wave. You can discount that the way you discount any analyst with a position.
Griffin has the opposite incentive. His business is built on the value of human judgment about markets. His most expensive line item is the salaries of people whose work he’s now watching get done by agents over a long weekend. If AI is real, his cost structure changes and his moat against everyone else with a Bloomberg terminal gets a lot thinner.
He had every reason to keep calling it garbage. He didn’t. He went on record at Stanford and said he was depressed by what he’d seen. That’s the data point.
Four months. Garbage to fairly depressed. From a guy who has access to better proof points than anyone reading this does.
What he didn’t walk back
Griffin didn’t soften the work-replacement claim. He didn’t add “but of course you still need humans for X.” He didn’t tell the Stanford room that the AI agents had limits he could name.
He told them to be lifelong learners. That was his constructive note: “The success in your career will be defined as to whether or not you will be a lifelong learner or not, and AI will just make this all the more important.”
That’s the most honest thing a $50 billion operator can say to a room of MBA students whose career math just got rewritten. Their only defense is staying ahead of the curve themselves.
So
If Griffin is depressed about what AI does to a Citadel PhD’s monthly output, the 30-person services firm with three Excel jockeys and an AP clerk does not get to be casual about this. They are not better positioned than Citadel. They are worse positioned. They just haven’t been in the room he was in, looking at what he was looking at, with the staff to verify it isn’t a parlor trick. He has seen the math. Everyone else is going to see it on a lag.