In the week of issued U.S. patents dated April 21, 2026, Bank of America Corporation appears in a thin blockchain grant column. The week produced 81 blockchain-tagged grants in total, the largest single bucket of which carried no named assignee; Bank of America, IBM, and nChain were the recognizable names with multiple records each. A granted patent is enforceable coverage of the claimed method, distinct from a published application that only discloses intent. Bank of America's entry, and the estate behind it, point at a specific layer: moving value and data between parties on a blockchain rather than at a consumer crypto product.

The grant is US12609917B2, “System and method for requesting data transfers in a blockchain network,” assigned to Bank of America Corporation and naming inventors Naga Vamsi Krishna Akkapeddi and Siten Sanghvi. The claimed method handles a request to move a quantity of “data objects” from one file to another by minting and exchanging digital tokens of equal value, classified in CPC H04L 9/50 (the blockchain-specific cryptography class), H04L 9/3213, and H04L 63/0428. The abstract describes the mechanism in plain sequence:

A processor receives a request to receive a transfer of a first amount of data objects into a first data file from a second data file. In response, the processor generates and transmits to the second data file a plurality of digital tokens amounting in value to the first amount of data object.— System and method for requesting data transfers in a blockchain network, US12609917B2

For a general reader, the business translation is straightforward: this is a token-settled transfer between two accounts or files, where the tokens stand in for the value being moved and the blockchain network records that the exchange completed. It is settlement plumbing expressed in on-chain terms, not a customer-facing wallet or exchange feature.

A recent, concentrated on-chain estate

The grant sits on top of a Bank of America blockchain estate that the records show is both large and recent. A search of issued grants assigned to the bank that reference blockchain returns 231 records, and the year breakdown indicates the build-out is concentrated in the last several years — 16 such grants dated 2019, rising to 42 in 2023 and 52 in 2024. The leading CPC class across that body is H04L 9/50 (50 grants), the blockchain-specific bucket, followed by hash and consensus classes such as H04L 9/0637 and H04L 9/3247.

Other recent grants extend the same payments-and-infrastructure theme. US12646044B2 (“Custom payment tokens for payments by using optical tones and neuromorphic spiking neural networks,” issued June 2, 2026) claims generating custom payment tokens validated through blockchain oracles and smart contracts. US12640947B2 (“Scalable cluster based scoring with multi-queue dynamic scheduling consensus mechanism in blockchain”) covers a consensus design intended to replace Proof of Work and Proof of Stake, classified in H04L 9/50. And US12639154B2 (“Auto-healing for blockchain configuration drifts”) addresses keeping the configuration of blockchain nodes consistent across a network — operational tooling for running a chain, not a token product.

What the claims add to the abstract

The independent claim makes the "data object" transfer concrete and adds detail the abstract omits. In claim 1, the blockchain network stores "a first data file associated with a first user and a second data file associated with a second user," and on a transfer request the processor generates "a first plurality of digital tokens, wherein each digital token represents at least one data object" and "amounts to the first amount of data objects to be received." Crucially, the tokens carry a payload of metadata: the processor transmits "information associated with the first plurality of digital tokens, wherein the information at least comprises an identity of the first user, an identity of the first data file, a reason of the requested transfer, and a due date when the requested transfer is due." Receipt of the tokens back at the first file is itself "a digital record of the transfer," and that receipt is what makes the system "determine that the requested data transfer has been completed." The tokens are not just value carriers — they are the settlement receipt.

The dependent claims turn this into recurring and multi-party settlement, governed by smart contracts. Claim 2 stores "a plurality of transfer rules" in "a first smart contract associated with the first data file." Claim 3 adds "a first transfer rule defining a frequency," so the contract regenerates and retransmits tokens "at each instance of the frequency" — a standing-order mechanism. Claim 4 handles fan-in from multiple counterparties, generating a "second plurality of digital tokens" for "a third data file associated with a third user," and claim 5 lets the system aggregate the inflows into "a third amount of data objects" forwarded "to a fourth data file." Claim 6 covers a chained arrangement where one counterparty must itself first collect "a first portion" from a third file before settling. Across system, method, and computer-readable-medium forms, the claims describe scheduled and multi-leg token-settled transfers rather than a single one-off exchange — the kind of standing settlement logic a bank would encode in a smart contract.

What the coverage maps to

Read together, the cluster traces a footprint at the level of operating and settling on a blockchain network: token-mediated transfers (US12609917B2), payment-token generation (US12646044B2), consensus mechanics (US12640947B2), and node operations (US12639154B2). That is consistent with a large bank documenting how it would move and settle value over distributed-ledger rails under its own controls, a pattern that recurs across other listed banks in the same weekly column. The volume is also notable against the weekly field: in a week of 81 blockchain-tagged grants, a major U.S. bank holding one of them while sitting on a 231-record estate is a meaningful share of a small set.

The standard caveats apply. The week's blockchain field is thin and dominated by unassigned records, so any single-week read is narrow. And a grant establishes that a claim issued, not how broadly it reads or whether the method is in production. What the records do show is the direction of Bank of America's accumulating coverage — token-settled transfers, consensus, and node operations on blockchain networks — with the April grant squarely inside that estate.