Introduction
Decentralized identifiers (DIDs) let the controller of a DID prove control of it without permission from any other party. But DID systems do not all have the same properties. The system that you anchor to sets the trust model that you get. Each option has tradeoffs.
Bitcoin has proof-of-work consensus, an immutable transaction history, and no central operator. These properties make Bitcoin a good foundation for some kinds of identity systems. Bitcoin is not the only valid approach.
Why the Anchor Layer Matters
The trust in a DID depends on the system that records its creation and its updates. The properties of the base layer set what an identity system on top of it can do. These properties include the order of operations, immutability, availability, and cost.
Bitcoin has these properties that are important for identity infrastructure:
- Immutability: To reverse a confirmed transaction, an attacker must redo the proof-of-work of that block and of all later blocks. If honest nodes control most of the CPU power, the probability of success decreases exponentially with each new block. DID operations that you anchor to Bitcoin get this finality.
- Censorship resistance: Many independent miners collect transactions into blocks. If one miner does not include a valid DID operation, another miner can include it in a later block. This is important when identity systems serve people in adversarial environments.
- Decentralization: No formal or informal decision body governs Bitcoin development. A consensus rule changes only when the Bitcoin ecosystem adopts and enforces the change. The locations of miners are less diverse. In a 2025 Cambridge survey, miners reported 75.4% of their hashrate in the United States, from a sample with a U.S. bias.
- Longevity: The Bitcoin network started in January 2009, and it still operates. Many identity systems must operate longer than the organizations that create them.
What It Costs
All technologies have costs. The costs of Bitcoin are real, and it is important to know them with current data.
Transaction fees change with network demand. The did:btc specification says that the price of block space "fluctuates greatly". When fees are high, one DID operation can be expensive. A batch puts many DID operations into one Bitcoin transaction, so the operations share one fee. Sidetree, which ION uses, allows up to 10,000 operations in one batch. The did:btcr2 Aggregate Beacons combine updates from many DID controllers into one Bitcoin transaction.
Confirmation latency is about 10 minutes per block on average. DID updates are not instant. For many identity use cases, this delay is acceptable, because you do not rotate keys every second. But it is a real constraint when you compare Bitcoin with systems that have finality in less than one second.
A full node stores the full blockchain and must stay in sync with the network. In September 2026, Bitcoin Core developers set the expected disk space for a mainnet node to 897 GiB. This value includes a margin of 5-10%. The value was 810 GiB in Bitcoin Core 30.0 (October 2025), an increase of 87 GiB in 11 months. Thus we recommend a drive of at least 2 TB.
The other hardware requirements are low. The bitcoin.org guide lists 2 GB of RAM and a broadband connection with an upload speed of at least 400 kilobits per second. umbrelOS runs on a Raspberry Pi 5, and MyNode software runs on a PC or a Raspberry Pi. As of September 2026, the Umbrel Home device costs USD 599 to 949, with 1 TB to 4 TB of SSD storage. The MyNode device has a 2 TB NVMe SSD.
Energy consumption from proof-of-work mining is a frequent criticism of Bitcoin. The Cambridge Digital Mining Industry Report (April 2025) estimates that Bitcoin mining used about 138 TWh per year. That is about 0.54% of global electricity consumption, with data from mid-2024. As of September 2026, the Cambridge Bitcoin Electricity Consumption Index (CBECI) gives a best estimate of about 152 TWh per year. Its range is 78 to 292 TWh. These comparisons give the numbers scale:
- In a typical US home, devices in standby mode use 5-10% of residential electricity (data from Lawrence Berkeley National Laboratory).
- In 2020, US homes used 190.6 trillion Btu of electricity for clothes dryers (EIA, RECS 2020, Table CE5.1a). That is about 56 TWh, for one appliance in one country.
- The Cambridge report found that surveyed miners used 52.4% sustainable energy: 42.6% renewables and 9.8% nuclear (data as of June 2024). Natural gas was the largest single source, at 38.2%. A location-based estimate in the report gives 47.9% sustainable energy, and the CBECI model gives 37.6%. The report says that the CBECI model uses location data from January 2022, and that the locations of miners changed after that date.
Energy use is a valid concern. We recommend that you use current data and compare it with other large uses of electricity.
How Bitcoin-Anchored DIDs Work
Many Bitcoin-anchored DID methods do not store full DID documents on-chain, because block space is expensive. Some methods, for example did:btc and did:btco, store DID data on-chain. The other methods use layers:
- Batch anchors: One Bitcoin transaction can commit to a hash of many DID operations. The operations then share the cost of that transaction.
- Off-chain document management: DID documents and operation data stay off-chain. Each method stores this data in a different way. ION uses content-addressable storage (IPFS). In did:btcr2, the controller gives the update history directly to verifiers as JSON, through sidecar delivery.
- Deterministic resolution: Any party with the same inputs can replay the operations and get the same current state of the DID. This party does not need to trust a central authority.
Comparison of Anchor Layers
Each DID base layer is optimized for different properties.
Proof-of-stake chains have lower transaction costs and faster finality. For identity operations that must be cheap and fast, this is a real advantage. Evaluate these tradeoffs:
- Validator concentration: The economics of proof-of-stake can give a small number of large validators a large influence. Delegation models are different on each chain.
- Economic finality: Proof-of-stake finality depends on economic incentives, not on energy expenditure. Your threat model and your time horizon decide if this finality is sufficient.
- Governance flexibility: Some chains can change consensus rules through governance. This ability is a feature for some use cases and a risk for others. For long-lived identity infrastructure, stability is important.
Non-blockchain approaches have a different set of tradeoffs. One example is the Mainline DHT of BitTorrent (BEP 5). It has high availability and no transaction costs, and it does not use proof-of-work. It does not give the strong order and immutability of a blockchain. For example, BEP 44 data can expire after 2 hours if nobody announces it again.
For identity systems where availability and cost are more important than an anchored history, DHT-based methods can be a good fit.
Lightweight methods such as did:key and did:web do not use a blockchain. A did:key DID comes from a public key, with no registry, so creation is instant and free. But did:key does not support updates, deactivation, or key rotation. The did:web method uses the DNS and HTTPS infrastructure of the web, and the domain owner controls the DID document. Neither method gives an anchored, immutable history, but many use cases do not need one.
The correct choice depends on the properties that your identity system needs.
Conclusion
Bitcoin gives decentralized identity real strengths: immutability, censorship resistance, and a long operational record. Its costs (fees, latency, and the energy for proof-of-work) are real. Node hardware is not expensive, and current Cambridge data gives the energy cost a measurable scale. Other approaches have a different balance of properties. The important question is not which base layer is better in general. The important question is which tradeoffs are acceptable for a given use case.