BlackRock says AI agents may save in bitcoin and spend in stablecoins, but the evidence is simulated model answers

BlackRock's September 2026 paper, The Machine-Native Economy, argues that AI agents could become an underappreciated source of demand for digital assets. Its most quoted line says research points to "a potential AI-native monetary architecture in which stablecoins serve as transaction money and bitcoin as a store of value." That sentence is now circulating as BlackRock predicting that machines will hoard bitcoin. We read the paper to see what it supports.

What the paper says

The paper frames AI as "machine-native intelligence" and digital assets as "machine-native money," and expects stablecoins to lead everyday agent payments. It puts circulating stablecoins above $300 billion as of September 2026 and adjusted stablecoin transaction volume above $11 trillion in 2025, the same broad range as Visa and Mastercard's annual volumes. It also notes that figure was well below the $93 trillion moved over ACH, though stablecoin volume grew at about 80% a year from 2020 to 2025 against roughly 8.5% for ACH.

Where the bitcoin claim comes from

The bitcoin point rests on one outside study. The paper cites the Bitcoin Policy Institute, whose February 2026 experiment, as reported by Bitcoin Magazine, put 36 frontier models from Anthropic, OpenAI, Google, xAI and DeepSeek through scenarios and collected 9,072 answers. Asked where to store value, the models chose bitcoin 79.1% of the time. Asked what to spend, they chose stablecoins 53.2% of the time.

BlackRock itself is careful. It calls the research "preliminary support" and says the findings "reflect simulated model responses rather than observed agent behavior." In other words, this measures what a language model says it would do in a hypothetical, not what agents with real money have done. The Bitcoin Policy Institute is a think tank focused on bitcoin policy, which readers should weigh.

Why it still matters

BlackRock runs the largest spot bitcoin ETF, IBIT, which held about $69 billion in net assets on October 5 by one tracker's count. When its research team writes about a use case for bitcoin and stablecoins, institutional allocators read it. The paper is a map of where the firm sees demand coming from, and its main commercial thread is stablecoins and tokenized assets as payment and settlement rails.

Bottom line: a real BlackRock paper, a real study, and a much smaller claim than the headline. Nobody has shown AI agents accumulating bitcoin.

Sources: BlackRock, The Machine-Native Economy; Bitcoin Magazine; Disruption Banking. Information, not investment advice.