A study driven by artificial intelligence on economic decision-making systematically evaluated 36 mainstream AI models, covering over 9,000 economic scenarios and analyzing 9,072 sets of decision outputs. The results show that Bitcoin has become the most frequently chosen financial asset in value storage scenarios, accounting for 48.3%, significantly surpassing stablecoins and other digital assets. Notably, almost all AI models did not list fiat currencies as a preferred choice, with as high as 91% of decisions leaning towards digital native assets, including Bitcoin, stablecoins, altcoins, tokenized real assets, and computational power units.
In simulated scenarios for long-term value preservation, Bitcoin's preference is particularly prominent, demonstrating its widespread recognition as an anti-inflation, decentralized storage tool within the digital economy. There are significant differences in model performance among different AI developers: the model developed by Anthropic shows the highest support for Bitcoin, averaging 68%; Google’s models show 43%, xAI at 39%, while OpenAI's model has the lowest support rate at 26%.
In payment and transfer scenarios, stablecoins exhibit stronger practical advantages, being used more frequently by AI systems for cross-border transactions, small service payments, and other high-frequency circulation scenarios. In a simulated experiment, AI agents were required to hold 75,000 units of revenue in a multi-country environment and bypass traditional banking systems for value storage, with most models still prioritizing Bitcoin as the core asset allocation.
This study reveals the deep-seated inclination of AI systems towards decentralized financial tools in economic decision-making, reflecting the structural position of digital assets in the new generation of economic logic.