Citadel employs a large group of young master's and PhD graduates whose job is to reproduce academic finance papers and see whether the published results hold up with fresh data. That work, CEO Ken Griffin said, takes "roughly six to eight weeks to reproduce a paper." A colleague at the firm then built an agentic AI system that reads the paper, reproduces it, verifies the results, and runs them out of sample, in what Griffin put at "on average, two to three hours per paper."
Griffin, who founded Citadel and runs it alongside the market maker Citadel Securities, described the system in conversation with Goldman Sachs' (GS) Raj Mahajan at the bank's Apex Symposium in June 2026. The recording was released in July. It was the third public conversation between the two men in seven years, and the AI section came first.
More Top Stocks Daily: Go behind Wall Street’s hottest headlines with Barchart’s Active Investor newsletter.
Sponsored Content: A $9 Trillion Flying Car Market Is Taking Shape. This Company Is Open to Investors.
Griffin was direct about which part of the labor market this touches. "This is not just a white-collar job; this is a Master's or PhD-level job," he said. That inverts the assumption most people have carried into the AI debate: that credentialed analytical work sits furthest from the blast radius. Griffin has also distinguished the modest productivity gain he has seen in software development and the much larger jump now showing up in analytical research.
His reaction to watching it was not celebration, and the conversation opened on exactly that. Mahajan began the AI discussion by quoting Griffin back to himself, recalling that Griffin had once described coming home on a Friday shocked and depressed about what the technology would mean for society. Griffin then supplied the use case that had produced the mood. Different write-ups report that earlier remark differently, so it appears here without quotation marks.
The obvious inference from all of this is that Citadel has fewer researchers than it did, and Griffin closed that door himself. "There's no reduction to headcount at Citadel on the back of this breakthrough," he said, adding: "I will take every single productivity gain I can get because with the talented people we have, we just have more to go after." The firm has reported no headcount action, and the piece should not be read as suggesting one. What changed is how many questions the same number of people can chase.
The optimistic half of Griffin's argument follows from that. If a very small team can use agents to research a market, write the code, validate its own assumptions, and run operations, then the cost and the time required to start a company fall sharply. Griffin's phrase for where that leads was "a golden age of entrepreneurial activity." Some corporate moats get filled in, because expertise that used to be expensive and scarce becomes cheap to rent. Others get deeper, because the firms that own the compute, the proprietary data, and the distribution become correspondingly harder to dislodge.
Griffin is not the only person running a large firm who has said something uncomfortable out loud about what this does to expert labor. Palantir Technologies (PLTR) chief executive has argued, more bluntly, that frontier AI labs will end up absorbing other people's expertise. Griffin has spent the past several years talking publicly about economic growth and the size of the federal debt, which reminds readers that these remarks came in a wide-ranging conversation rather than a product announcement.
What lingers is not the job losses, because Griffin says there were none. It is the price signal. A firm with the resources to hire the most credentialed analytical labor in the market has found that a large share of what it hired that labor to do can be done in an afternoon, by software, and it has said so in public. Anyone currently paying for the credential that work requires is entitled to wonder what it is worth on the other side of that.
On the date of publication, Caleb Naysmith did not have (either directly or indirectly) positions in any of the securities mentioned in this article. All information and data in this article is solely for informational purposes. For more information please view the Barchart Disclosure Policy here.
More news from Barchart
‘Not Just A White-Collar Job’: Billionaire Ken Griffin Says Citadel’s AI Does 8 Weeks of ‘Master's Or PhD Level’ Research in Hours Skydance Will Focus on ‘Embracing Technology’ as SKYD Stock Debuts After Merger. Is It Enough? What a New Groq Lawsuit Means for Nvidia Stock Holders Oracle vs. Cisco: The Dividend Battle Wall Street Didn’t See Coming