Poor Man's Light Clients for Cardano

Trust-minimized blockchain access via Merkle proofs

The Challenge: Full Nodes Are Heavy

Today's Reality

  • Running a Cardano full node requires ~300GB of storage
  • High memory requirements
  • Constant network availability and bandwidth

Tomorrow's Reality

  • Storage requirements will only grow
  • Hardware requirements will rise with new features
  • Amaru will help a bit, but not enough

This kills participation

Why Does This Matter?

Decentralization depends on accessibility

  • Mobile wallets can't run full nodes
  • Browsers can't run full nodes
  • IoT devices can't run full nodes
  • Developing countries face infrastructure barriers

Result: Most users depend on centralized services for blockchain data and transaction building.

The Trust Problem

Application servers provide blockchain data and unsigned transactions for users to interact with the network.

  • Without a full node, users must trust the server
  • Trusting a centralized server undermines decentralization
  • Even honest servers bring risks: hackers can compromise them
  • Users have no way to verify data integrity

Current Light Client Options

Approach Trust Required
Centralized API Trust the provider completely
Full node Trust no one (but heavy)

We need a middle ground

We Can Prove an Untrusted Server Is Honest

Cryptographic Proofs Remove Trust Assumptions from servers of web3 applications.

If untrusted servers provide proofs that clients can verify independently, we remove the need to trust them:

  • Good for the server: they're protected even if hacked or compromised
  • Good for the client: they trust math, not promises
  • Good for trusted parties: they build reputation without involvement in specific applications

The Solution: Merkle Trees

A Fingerprint for the Entire UTxO Set

        Root Hash (32 bytes)
       /                    \
    Hash                    Hash
   /    \                  /    \
 UTxO  UTxO              UTxO  UTxO
  • A single root hash represents millions of UTxOs
  • Anyone can verify that a UTxO belongs to the set
  • Changing any UTxO changes the root

The Simplest Possible Infrastructure

Publishing a Merkle root is trivial

A JSON file with UTxO set root hashes covering the rollback window, updated every block (~20 seconds):

[
  { "slot": 123456789, "root": "a1b2c3..." },
  { "slot": 123456769, "root": "d4e5f6..." },
  { "slot": 123456749, "root": "g7h8i9..." },
  ...
]

Can be hosted anywhere:

  • GitHub release asset
  • Google Drive
  • S3 bucket
  • Even a tweet/X post

Poor Man's Multisignature

Multiple independent publishers = distributed trust

Light clients sample 2-3 sources: match → verified | differ → investigate

Who Should Publish?

Institutions (low effort, high impact)

Publisher Why
Cardano Foundation Ecosystem steward
IOG Core developers
Emurgo Enterprise adoption
Mithril team Already running infra

Anyone else

  • Stake pool operators
  • Exchanges
  • dApp developers
  • Community members
  • Any company using Cardano for business operations

More publishers = stronger decentralization

The Architecture

Benefits Summary

Stakeholder Benefit
End Users Fast, verifiable access to dApps
Developers Simple API, no node required, just proof verification libraries
Ecosystem More decentralized infrastructure
Publishers Low effort, high impact reputation boost

Cost to Participate

Role Infrastructure Interface
Root Publisher Node + cardano-utxo-csmt Static file hosting (GitHub, S3, etc.)
Application Server Node + cardano-utxo-csmt Patch API to serve proofs with data
Light Client None Proof verification library (browser/mobile)

Publishing roots costs almost nothing

Current Status

Working Today

  • ✅ Merkle tree construction from UTxO set
  • ✅ Proof generation and verification
  • ✅ HTTP API for light clients
  • ✅ Fast bootstrap via Mithril
  • ✅ Real-time chain sync

Next Steps

  • Integration with wallet SDKs
  • Public testnet deployment
  • Performance optimization

The Vision

A Cardano where:

  • Application servers don't bear the burden of trust
  • Light clients choose their trusted publishers from multiple independent institutions

Questions?

Resources

  • GitHub: lambdasistemi/cardano-utxo-csmt
  • Documentation: /docs