Ken Griffin: Citadel's AI System Completes 8-Week PhD-Level Research in 2 Hours, Ushering in a Golden Era for Startups
Billionaire and Citadel founder Ken Griffin stated that an internal AI system at the company replicated a PhD-level financial research task that typically takes 6 to 8 weeks in just 2 hours. He believes this will trigger a "golden era" for entrepreneurship.
At a Goldman Sachs forum, Griffin explained that Citadel's team, previously composed of master's and PhD researchers, spent about 6 to 8 weeks reproducing academic papers: reading hypotheses, reconstructing methods, validating results, and conducting out-of-sample tests. The AI system developed by colleagues now averages only 2 to 3 hours to complete the entire process.
He emphasized that this work is at the master's or PhD level, not typical white-collar tasks. Citadel has not cut any jobs as a result, but instead has used the productivity gains to tackle more problems. Griffin believes AI is rapidly leveling the competitive playing field, allowing small teams to challenge traditional giants.
From a market mechanism perspective, the view reinforces expectations of AI's impact on knowledge work and the barriers to entrepreneurship, with funding flowing toward AI systems and efficient startup tools. Beneficiaries will be entrepreneurs and quantitative firms that effectively utilize AI, while traditional models relying on large research teams will face pressure.
Source: Public Information
ABAB AI Insight
Ken Griffin's Citadel is known for its quantitative and data-driven approach, relying on highly educated talent to reproduce and validate academic results to find trading signals. The internal AI agent has compressed the research cycle from weeks to hours, demonstrating breakthroughs in complex multi-step knowledge work.
In terms of capital pathways, Griffin interprets the efficiency gains as entrepreneurial opportunities rather than mere signals for layoffs, emphasizing that small teams can match or exceed the capabilities of large organizations with AI, strategically encouraging the use of technology to lower the barriers to entrepreneurship.
Similar cases can be seen in other hedge funds and tech companies reporting AI accelerating research or code generation, along with discussions about the rapid erosion of competitive moats by technology, currently in a phase where knowledge-intensive work is significantly accelerated by AI systems.
The structural judgment indicates technological substitution: agent-based AI is replacing parts of high-skill research processes, automating multi-step verification work, thereby reducing the reliance of entrepreneurship and innovation on large-scale manpower and accelerating market competition.
ABAB News · Cognitive Laws
- PhD-level work can be compressed to hours
- Filling moats creates entrepreneurial windows
- Productivity gains expand the boundaries of problems