BlackRock Maps the Convergence of AI and Digital Assets
BlackRock has published a research note arguing that artificial intelligence and digital assets are converging around a shared foundation, laying the groundwork for what it calls a “machine-native economy.” The asset manager’s analysis points to large language models and blockchain-based tokenization as complementary technologies, with stablecoins emerging as a natural payment rail for autonomous software agents and compute capacity shaping up as a new digital asset market.
The Shared Foundation: Data, Verification, and Settlement
At the core of the thesis is a simple observation: both large language models and public blockchains depend on vast, verifiable datasets and deterministic settlement. LLMs need structured, high-quality data to train and reason; tokenized networks need reliable records of ownership and transfer. When those two systems interact, machines can not only transact but also verify the terms of those transactions without human intermediation.
BlackRock’s researchers frame tokenization as the bridge. Real-world assets, from treasuries to private credit, become programmable objects that AI agents can allocate, collateralize, or rebalance. Stablecoins, meanwhile, provide the unit of account and medium of exchange that keeps settlement instant and borderless — a requirement for software that operates continuously rather than during banking hours.
Compute as an Emerging Asset Class
Another pillar of the report is the idea that compute itself is becoming a tradable digital asset. As AI training and inference demand surges, GPU time and data-center capacity are being financialized through tokenized markets, futures, and decentralized compute networks. BlackRock’s framing suggests this could evolve into a distinct market segment, with its own pricing dynamics, hedging instruments, and risk profiles — much like commodities or bandwidth before it.
- Stablecoins as machine money: programmable dollars enable agent-to-agent payments without human sign-off.
- Tokenized collateral: AI-driven portfolios can post and move tokenized assets in real time.
- Compute markets: GPU capacity and inference may become a new investable digital asset class.
Implications for Institutions
For institutional investors, the report implies that AI and digital-asset strategies should no longer be siloed. A fund building AI-driven trading infrastructure will increasingly need on-chain settlement and tokenized collateral to function at machine speed. Custodians, exchanges, and asset managers that treat the two trends separately risk building systems that cannot interoperate.
The regulatory picture remains unsettled. Stablecoin legislation, tokenization rules, and oversight of decentralized compute networks are all still evolving, and any of them could slow the convergence. But BlackRock’s core argument is directional: the infrastructure for a machine-native economy is being assembled now, and the firms that connect AI to tokenized rails early will define how it operates.
Forward Look
The next phase will likely be defined by pilots rather than proclamations — tokenized money-market funds used as collateral by AI allocators, stablecoin rails tested for agent payments, and compute-backed instruments listed on regulated venues. If those experiments hold, the boundary between AI infrastructure and digital-asset markets may blur faster than most institutions expect.
