Citadel CEO Ken Griffin: Artificial Intelligence is Almost Certainly in a Bubble
Hedge fund Citadel's founder and CEO Ken Griffin stated at the company's annual Global Markets Future Conference that investors are almost certainly in an artificial intelligence bubble.
This statement came during an interview with Bloomberg, where Griffin pointed out that any fundamental economic transformation is often intertwined with bubbles; he previously compared the current situation to echoes of the internet bubble but emphasized that true winners and losers are difficult to identify in the early stages.
Earlier this year in Davos, Griffin referred to the hype around artificial intelligence as a means to support massive expenditures, noting that U.S. data center spending is expected to exceed $500 billion this year; he also stated that generative tools still fail under deep testing, but recently acknowledged that AI has become real and powerful.
He described how Citadel has deployed intelligent systems that can replicate, validate, and conduct out-of-sample testing of academic finance papers that would typically take a master's or PhD team 6 to 8 weeks to complete, reducing the time to just 2 to 3 hours; this experience left him feeling frustrated, as high-skilled knowledge work could be rapidly automated.
Griffin also believes that AI will usher in a golden age of entrepreneurship, lowering barriers to entry and enabling small teams to challenge existing companies more quickly; however, he pointed out that rising computing costs have created new moats for companies with large capital.
Citadel manages approximately $68 billion, long utilizing machine learning technology; Griffin's personal net worth is about $48 billion, with holdings in AI-related companies such as Nvidia, Amazon, and Microsoft.
In market mechanisms, bubble rhetoric serves as a signal of public market sentiment, potentially affecting the flow of funds into overvalued AI stocks and infrastructure; beneficiaries may be companies with actual productive applications, while those under pressure are startups and hardware expansion projects whose valuations rely mainly on narratives rather than realized earnings.
Source: Public Information
ABAB AI Insight
Ken Griffin has long managed Citadel and has historically taken a cautious stance on artificial intelligence, previously stating in Davos that generative tools are mostly "garbage" after in-depth testing and noting that data center spending needs to be supported by grand narratives; recently, however, he has shifted to acknowledge that AI is real and usable, deploying intelligent systems internally to significantly compress research cycles.
The capital path is Citadel's own use of AI to accelerate internal research, while Griffin personally and the company hold core positions in Nvidia, Amazon, Microsoft, and other key stocks; the motivation is to leverage computing power and models to enhance investment research efficiency while remaining vigilant about valuations in the public market that overly rely on future narratives.
This is similar to the judgments of some hedge fund managers around the year 2000 regarding the internet bubble, acknowledging that the direction of technological change is correct while warning that valuations may deviate from fundamentals; the current AI field is in a phase of large-scale capital expenditure and actual productivity validation, with some applications already implemented, but overall spending and expectations still significantly outpacing realized returns.
The essence lies in changes in pricing power and capital concentration: massive data center and computing expenditures will concentrate supply on a few companies capable of financing and providing power, with the mechanism being that after narrative-driven capital inflows, only participants who can translate technology into quantifiable productivity will survive after the bubble bursts, shifting valuations from future stories to realized efficiencies.
ABAB News · Cognitive Law
- The correct direction of technology does not mean every valuation is correct.
- Bubbles often accompany true transformation, but most participants will still disappear.
- Expenditures need narrative support; when narratives fail, only balance sheets remain.