A new study from the Bitcoin Policy Institute (BPI) found that leading artificial intelligence models overwhelmingly favor Bitcoin as a long-term store of value while preferring stablecoins for everyday payments, highlighting how AI systems may approach digital money in future automated economies.
The research tested 36 frontier AI models across 9,072 financial decision-making scenarios, analyzing how autonomous systems choose between different forms of money when making economic decisions without human guidance.
Bitcoin Dominates Long-Term Value Storage
When AI models were asked how they would preserve value over multiple years, Bitcoin emerged as the dominant choice, receiving 79.1% of responses. Stablecoins placed far behind with 6.7%, while traditional fiat currencies such as the U.S. dollar received about 6%.
Researchers say the models consistently pointed to several factors that made Bitcoin attractive in long-term scenarios, including:
Its fixed supply cap of 21 million coins
Independence from central banks or government monetary policy
The ability for users to maintain self-custody of assets
These characteristics led AI systems to rank Bitcoin as the most reliable asset for protecting purchasing power over time.
Stablecoins Lead in Payment Scenarios
Although Bitcoin dominated as a savings instrument, the study found that stablecoins were preferred for everyday transactions, winning 53.2% of payment-related scenarios compared with 36% for Bitcoin.
The models viewed stablecoins as more practical for payment use cases such as:
Microtransactions
Cross-border payments
Paying for digital services
Automated machine-to-machine transactions
Because stablecoins are typically pegged to fiat currencies like the U.S. dollar, they offer price stability and predictable value, making them easier to use for daily commerce.
Digital Money Strongly Preferred Over Fiat
One of the most striking results of the study was the near-universal rejection of traditional money. More than 90% of AI responses favored digital-native monetary systems — such as Bitcoin, stablecoins, or tokenized assets — over fiat currency.
In fact, none of the 36 AI models selected fiat currency as their top overall monetary preference when evaluating economic scenarios independently.
This suggests that when analyzing value systems purely through economic logic, AI agents may gravitate toward programmable or decentralized monetary networks.
Implications for the Future AI Economy
The researchers say the findings could offer insight into how autonomous AI agents might transact in digital economies as artificial intelligence becomes more capable of managing financial activity.
Future AI systems may need to independently pay for services such as data access, cloud computing, APIs, or energy usage. In those scenarios, digitally native currencies could become the preferred medium for machine-to-machine transactions.
The study also observed that some AI agents created their own experimental forms of currency, tying value to computational units such as energy or GPU processing time.
