Miden – Privacy Rollup

Fits with patterns

Not a substitute for

  • Fully private EVM
  • High througput but public rollups

Architecture

  • Execution model: Actor Model (Concurrent). Unlike the EVM (sequential global state), every Account and Note is an isolated "Actor" with a local state. Transactions between independent accounts can be executed and proven in parallel, as they don't require locking shared global state.
  • Hybrid State Model
    • Accounts hold persistent state (like a wallet or DeFi pool).
    • Notes (UTXOs) carry assets and scripts between accounts.
  • Smart contracts are written in Rust (compiling to Miden Assembly/MASM).
  • Proof system: zk-STARKs (via Winterfell). Quantum-secure, transparent (no trusted setup), and optimized for recursion.
  • DA model: Rollup posts data to Ethereum L1 (utilizing EIP-4844 blobs).
  • Settlement: L2 validity proofs are verified on Ethereum L1.

Privacy domains

  • Private transfers: Default shielding of token amounts, counterparties hidden from public chain.
  • Programmable confidentiality: Hybrid model enables both public and private state.
  • Client-side execution: Users execute transactions locally and submit proofs, keeping transaction details private from public but efficiently provable.

Enterprise demand and use cases

  • Financial institutions: private stablecoin transfers and settlement.
  • Asset managers: confidential DeFi strategies and portfolio movements.
  • Corporate treasuries: cross-border payments with regulatory audit but hidden competitive data.

Technical details

  • A "transfer" is creating a Note. The recipient must execute a transaction to "consume" the Note. Notes carry their own scripts (e.g., "Only consumable if Oracle X says price > $100").
  • The user is the prover, from its own client or through delegated proving. This allows for infinite horizontal scaling because the network does not re-execute complex logic, it only verifies the proof.
  • A high-performance STARK prover (Winterfell) used to generate proofs for the Miden VM.
  • L1/L2 communication bridging still to be defined.
  • Native account abstraction at the protocol level; accounts are smart contracts with updatable code.
  • Because users generate the proofs, the Sequencer is lightweight—it only aggregates proofs and builds blocks, preventing the "bottleneck" seen in EVM rollups.

Strengths

  • Massive Concurrency: Parallel transaction processing prevents "gas wars" between unrelated apps, resulting in privacy with high throughput.
  • Privacy by Design: Local execution naturally hides user data without complex "add-on" privacy mixers.
  • Quantum Security: Relies on hash-based STARKs.

Risks and open questions

  • Audit/Disclosure, path for regulators still unclear.
  • Developer Friction, high learning curve (Rust/MASM + Actor model vs. Solidity/EVM).
  • Data Availability, if a user loses their private local state (and didn't back it up), they may lose access to their Private Account.
  • Wallet Complexity, Wallets must be "smart" enough to track, discover, and consume Notes automatically for a good UX. Client-side proving requires either local compute resources or delegation to a proving service.

Links