AI Models Prefer Bitcoin, Study Finds
A new study by the Bitcoin Policy Institute found that frontier AI models prefer Bitcoin and digital money over traditional fiat when asked how they would transact as autonomous economic agents. The research tested 36 different AI models, suggesting a native preference for programmable, digital-first currencies.
The study's findings quantify an emerging consensus: autonomous agents will require digitally native, permissionless money. Over 9,072 scenarios, not a single one of the 36 AI models tested chose a traditional fiat currency as its top preference. Instead, they overwhelmingly favored digital assets, with Bitcoin selected in 48.3% of all responses and stablecoins in 33.2%. This points to a foundational architectural alignment between AI and crypto rails. Diving deeper, the models demonstrated a sophisticated, two-tiered view of digital money that mirrors established financial principles. For long-term value preservation, Bitcoin was the undisputed choice, selected in 79.1% of store-of-value scenarios. For transactional purposes, however, stablecoins were preferred, leading in 53.2% of payment scenarios, indicating a clear distinction between a primary reserve asset and a medium of exchange. This preference is already being met with infrastructure. Tether's co-founder, Reeve Collins, has called AI the "browser moment" for crypto, envisioning a future where conversational AI agents manage all user transactions. Major industry players are building for this reality; Coinbase has launched "Agentic Wallets," and Stripe's co-founder John Collison predicts a "torrent" of AI-driven commerce running on stablecoins. The "AI x Crypto" narrative is rapidly moving from a speculative thesis to a core investment area for venture capital. Crypto-native VC firm Paradigm is reportedly raising up to $1.5 billion to expand its investments into AI and robotics. This follows significant funding rounds for on-chain AI projects, including decentralized AI platforms and infrastructure designed to support an autonomous agent economy, signaling a capital rotation towards this converging sector. This trend is creating a distinct hunting ground for sourcing deals in "on-chain AI." Projects are emerging across the stack, from decentralized compute and AI model marketplaces like Cortex and Fetch.ai to protocols focused on AI-driven data verification and smart contract generation. Layer 1 blockchains are also competing to become the primary settlement layer for AI agents, with Solana launching an "AI Agent Registry" and Base fostering an AI economy around the x402 protocol. For institutional allocators, the convergence of AI and crypto represents a new industrial order. The thesis extends beyond simple automation, suggesting that blockchains will serve as the essential economic substrate for autonomous systems to transact and manage value. This creates opportunities not just in tokens but in the "picks and shovels" of the machine intelligence age—the data infrastructure, decentralized compute, and privacy-preserving technologies that will underpin this new economy.
