Banking & Capital Markets Systems · Chapter 12 of 13
The consortium graveyard, and the three ways it failed
Four of five trade-finance consortiums are dead — including one that proved the technology worked. The failure modes are not technical, and they generalise.
Between roughly 2017 and 2023, the world's largest banks and shipping lines built shared blockchain platforms for trade finance. Almost all of them are gone. This is the single best-documented natural experiment in enterprise blockchain, and it repays study far beyond this subject.
| Consortium | Built on | What happened |
|---|---|---|
| we.trade | Hyperledger Fabric | Insolvent, June 2022. Twelve European banks plus IBM. |
| Marco Polo | R3 Corda | Insolvent, early 2023. A roughly $12m Bank of America buyout fell through. |
| Contour | R3 Corda | Shut, November 2023. See below — this is the important one. |
| TradeLens | Hyperledger Fabric | Wound down, Q1 2023. Maersk's own words: it had not reached commercial viability. |
| Komgo | Quorum | Still operating and growing — 10,000+ corporate users, 60+ banks. |
Contour is the case that should change how you think about this.
It worked. Letter-of-credit processing fell from around ten days to under twenty-four hours — a 90% reduction, which is precisely the outcome the whole exercise was for.
It shut anyway, at 60–70 transactions a month commercially. Banks pulled funding.
A technology can deliver exactly the improvement it promised and still fail, because the improvement was never the binding constraint. If you take one thing from this book into an unrelated project, take that.
Three distinct failure modes
- 1. Governance. Competing bank-shareholders could not agree how to fund or run a shared platform. Everyone wanted the network; nobody wanted to pay for someone else's advantage in it.
- 2. Business model. Contour's per-transaction fees could never cover its costs at that volume. The stated reason customers did not come: the value proposition was too narrow to justify corporates changing their processes. A 90% cycle-time improvement on one document type is not enough to move an operations department.
- 3. Competitive trust. TradeLens was part-owned by Maersk — so Maersk's shipping competitors were never going to participate fully. A platform visibly controlled by one dominant player cannot get genuine buy-in from its rivals, whatever the ledger underneath does.
Read those against question 1 of the four-question test. All three are failures of getting mutually distrustful parties to actually co-write. None is a failure of cryptography, throughput or engineering. The technology was never the bottleneck, in any of the five.
India's own attempts
- IDRBT, RBI's research arm, ran India's first institutional proof of concept in January 2017.
- BankChain — started by SBI in February 2017 with around twenty-seven banks, for shared know-your-customer records. No evidence of activity after roughly 2019–20.
- IBDIC — fifteen major banks doing inland letters of credit on Infosys Finacle Connect, cutting cycle time from eight or nine days to two or three. Its invoice-financing system passed the RBI Regulatory Sandbox in 2025. FY25 revenue: ₹2.82 crore. It also renamed itself from Indian Banks' Blockchain Infrastructure Company to Indian Banks' Digital Infrastructure Company — dropping the word from its own name is a signal worth noticing.
- NPCI Vajra — announced January 2020 as a permissioned ledger for UPI and RuPay reconciliation. No evidence it reached production.
And then the part that reframes everything
Two Indian systems solved the same class of problem without a ledger
Account Aggregator — consent-based sharing of financial data between institutions that compete with each other. Textbook multi-party trust. It has passed 2.61 billion linked accounts, and it was built specifically without distributed ledger technology. CKYC is likewise a plain centralised registry — while BankChain, which set out to do shared KYC on a ledger, went quiet.
The Bills of Lading Act, 2025 (in force July 2025) gave electronic bills of lading the same legal standing as paper, aligned to the UNCITRAL model law, with a phased mandate — 25% electronic by 2027, 60% by 2029, all of it by 2031.
That legal-recognition gap is exactly what Contour's own post-mortem named as a root obstacle. India closed it with an ordinary statute.
The pattern in both: fix the standard and the governance, and the coordination problem yields — without changing the data structure at all.
What this means for your career, said plainly
There is no credible, dated source isolating blockchain-specific role counts in Indian BFSI. What turns up in a search is course marketing, and it is not fit to cite. For context, Indian technology employs about 5.80 million people and BFSI technology is projected at 2.4 million by 2030 — but that covers cloud, AI and core banking, with no blockchain breakout inside it.
The realistic Indian role is one of three: a systems-integrator job at TCS, Infosys or Wipro building a bank-commissioned pilot; a policy or compliance role at a regulator; or a payments engineering job where distributed ledgers are one skill among several. It is a skill to layer onto a BFSI or payments career — not yet a standalone bet, the way UPI integration already is.
Which is a genuinely useful position to hold. Being the person in the room who knows why Contour died, what atomic settlement actually removes, and what the e₹ numbers say is worth considerably more than being the person who can write a token contract — and far fewer people can do it.
Check yourself