CZ: Believes there is a certain paradox in AI trading agents, but is confident that AI will be widely adopted by traders

On December 18, Binance founder Changpeng Zhao (CZ) shared his thoughts on AI’s role in trading during an end-of-year Q&A session: “AI will become widely adopted in trading, but I see multiple potential development paths—ones that differ somewhat from prediction markets. Nearly all seasoned traders, large trading firms, and hedge funds have dedicated trading teams, and I suspect many are already using some form of AI, even unknowingly (for example, in data analysis). Serious players will either train their own AI algorithms or at least test AI trading platforms. AI trading platforms face a potential contradiction: If you have an AI algorithm that generates significant profits, why sell it as a service? Why not trade with it yourself? The only plausible reason might be a lack of capital—but today, fundraising is relatively easy for strong teams, especially those with a genuinely profitable AI. Selling AI as a monthly subscription implies the service is more profitable than trading with the AI directly. Inevitably, this means the AI’s earnings are less than total revenue from paying subscribers. A counterargument is that AI acts as a tool to simplify customizing a trading system (instead of building from scratch), with each user’s version slightly unique—but I find this argument somewhat weak and unconvincing. Moreover, a highly successful AI algorithm will lose effectiveness if adopted by too many traders. Markets are a game of group psychology—you’re essentially trading against the crowd. If everyone uses the same strategy, early adopters typically profit, but latecomers may not (even with the same strategy). Speed, performance, and other execution factors then become critical. Overall, I think AI will be widely adopted by traders in their own unique, customized ways.”