Enterprise & permissioned chains
When the participants are known, regulated companies, you don't need mining — you need identity, privacy between competitors, and governance that survives budget season. Meet Fabric, Corda, and Besu: the platforms that carried nearly every insurance consortium.
L1Why permissioned exists
Public chains solve trust among strangers with expensive consensus and total transparency. An insurance consortium has the opposite profile: members are identified and contracted, but they're competitors — they want shared workflows with selective visibility.
| Need | Public chain answer | Permissioned answer |
|---|---|---|
| Who can join / write? | Anyone with gas | Identified members via certificates (MSP) / network operator |
| Who sees my data? | Everyone, forever | Only counterparties (channels, private state, need-to-know) |
| Finality & throughput | Minutes-ish · ~15–100 TPS L1 | Instant BFT finality · 1,000s TPS (Fabric benchmarks ~3,500) |
| Compliance | Pseudonymous — bolt KYC on top | Identity-first by construction |
| Who pays / governs? | Fee market, open governance | The consortium — which is exactly where B3i and we.trade died |
L2Hyperledger Fabric — channels & endorsements
Fabric (Linux Foundation) is a modular consortium platform. Its two signature ideas:
- Chaincode = smart contracts in Go/JavaScript/Java — general-purpose languages your team already writes. You'll read a claims chaincode in Lab M07.
- Channels + private data collections = privacy by partition: competitors share the network, not the data.
- Used by: openIDL (regulatory reporting, AAIS), IBM's insurance samples, we.trade (RIP — governance, not tech).
L3R3 Corda — point-to-point by design
Corda drops the "global broadcast" idea entirely: there is no global ledger. Data moves only between transaction parties via flows; a notary cluster merely attests uniqueness (no double-spend) without seeing full contents. States are UTXO-like immutable facts ("Policy #123 v3") consumed and reissued.
- CorDapps in Kotlin/Java; legal-prose linkage was a founding feature (contract ↔ legal doc).
- Built for regulated finance from day one — R3 reported 20+ regulated networks and $17 B+ tokenized RWA live on Corda by 2025.
- Used by: Insurwave (marine), B3i's Cat XoL product, Marco Polo (RIP), many digital-bond platforms.
L4Besu / Quorum — the EVM, permissioned
Third path: keep Ethereum's tooling (Solidity, wallets, the entire M02 skill set) but run it as a permissioned network with IBFT consensus and optional private transactions. Hyperledger Besu is the leading client; JPMorgan's Quorum pioneered the pattern and powered early JPM Coin. Sweet spot: consortiums that want EVM talent portability and a future bridge to public L2s (exactly the Kinexys→Base trajectory in Module 09).
L5Choosing — comparison + live worksheet
| Fabric | Corda | Besu (perm. EVM) | Public L2 | |
|---|---|---|---|---|
| Privacy model | Channels + private collections (explicit design) | Point-to-point by default (strongest) | Privacy groups (Tessera) — added on | Public (zk tricks emerging) |
| Contract language | Go / JS / Java | Kotlin / Java | Solidity | Solidity |
| Finality | Instant (Raft/BFT) | Instant (notary) | Instant (IBFT) | Seconds + L1 settlement |
| Throughput | ~3,500 TPS benchmarked | High, per-flow | ~1,000s TPS | 100s–1,000s |
| Ecosystem/talent | Enterprise Java/Go shops | Banking-centric | Huge (EVM) | Huge (EVM) |
| Insurance track record | openIDL, IBM samples | Insurwave, B3i, bonds | Banking rails (Quorum lineage) | Etherisc, Arbol, Lemonade |
| Watch out for | Ops complexity (CAs, channels) | Smaller talent pool, R3 dependency | Privacy add-ons maturity | Data publicity, gas UX |
Worksheet 3.1 · Score your own use case
interactiveRate how much each factor matters for your scenario (0 = irrelevant, 3 = critical). The totals re-weight live. Defaults describe a 5-insurer motor-claims consortium.
| Factor (weight 0–3) | Weight | Fabric | Corda | Besu | Public L2 |
|---|---|---|---|---|---|
| Competitor-grade privacy | 3 | 3 | 2 | 0 | |
| EVM/Solidity talent reuse | 2 | 2 | 3 | 3 | |
| Fine-grained approval policies (who must sign) | 3 | 2 | 2 | 1 | |
| Regulated-finance track record | 2 | 3 | 2 | 1 | |
| Reach retail customers / public composability | 1 | 1 | 2 | 3 | |
| Minimal consortium ops burden | 1 | 1 | 2 | 3 | |
| Weighted total | — | — | — | — |
No single winner exists — the exercise is noticing which weights flip the answer. Present those weights, not the tool, to your steering committee.
L6Consortium governance — where projects actually die
What must be governed
Money — who funds build & run, forever? Operations — who runs nodes, upgrades, incidents? Data — who owns records, who may join/leave with what? Legal — is a chain record contractually binding? Which regulator has questions?
How B3i died (2022)
Twenty-plus insurers, working software on Corda, a live Cat XoL product — and insolvency anyway, when a funding round failed. Technology adequate; the money ring broke. Every consortium pitch you ever review should be tested against this case first (full autopsy in M08).
Galileo skipped consortium governance entirely by being single-operator — one company, three nodes, no committees. That's why it shipped in production while consortia stalled… and also why it earns none of the multi-party benefits. The capstone asks you to weigh exactly this trade.
▶Watch
Crisp whiteboard explainers on Fabric, blockchain-vs-database, consensus.
Fabric workshops, openIDL sessions, real deployment talks.
Sources: Hyperledger Fabric docs; R3 Corda docs & 2025 RWA figures (R3); Fabric TPS from published benchmarks; consortium histories per Ledger Insights / GTR reporting (as of Aug 2026).