Walled garden vs open ledger: will India's depository-custodied DLT bonds win institutions where public chains couldn't?

On September 10, at a fintech festival in Mumbai, India switched on what it is calling Demat 2.0: the first tokenized corporate bonds settled instantly against the Reserve Bank of India's digital rupee. Three issuers — state power lender REC, engineering group L&T, and IIFL Finance — sold roughly ₹1,025 crore (about $120 million) of three-year paper in a matter of days. The headlines read "India launches blockchain bonds." The reality is narrower and, in a way, more interesting: this is a walled garden, not an open ledger. Whether that walled garden is the future of tokenized debt, or a proof-of-concept that never escapes its corners, turns on a single number that won't exist until December.

First, a definition, because the word "tokenized" is doing heavy lifting here. In the crypto version you've probably read about, a bond is tokenized on a public chain like Ethereum: the token lives on a network anyone can join, anyone can build on, and — technically — anyone can trade. That openness is the selling point. It is also why institutions have mostly stayed away: the legal category of the token is ambiguous, custody is murky, and "who holds the register" is a question with no comfortable answer. Public-chain tokenized debt keeps working, but it stays a niche.

India's pilot is the mirror image. The bonds are created as digital tokens, but they live on a distributed ledger owned by the depositories, NSDL and CDSL — the same regulated institutions that hold every Indian stock in demat form today. No one is disintermediated. The depository keeps the authoritative register and remains the registered owner; the bank investor is the beneficial owner; a custodian bank holds the wallets. This is not a new asset class escaped into the wild. It is a faster, atomically-settled version of the existing depository system, wrapped in the word "blockchain."

Walk the custody chain actor by actor and you see the design intent. SEBI regulates the securities and launched the pilot. The RBI supplies the settlement asset — its own wholesale digital rupee, central-bank money rather than a private stablecoin. The depositories run the ledger and keep legal ownership. Participating banks hold the digital-rupee wallets for the beneficial owners. An interface the pilot calls the Unified Market Interface ties the two legs together so that neither the bond token nor the digital rupee moves unless both move at once — atomic delivery-versus-payment, which collapses the two-day settlement gap that leaves bond buyers exposed to counterparty risk.

That is a real engineering achievement, and it is exactly the pitch an institutional allocator wants to hear: legal certainty (the register is where the law already looks) plus custody comfort (regulated depositories and central-bank money) plus a concrete efficiency gain (instant, atomic settlement). The public-chain model offers none of those by default. On paper, the traded-offs of the walled garden look acceptable — you lose the open programmability and composability that let developers stack lending, derivatives, and collateral on top of a token, but the institutions this targets never wanted that composability anyway. They wanted the register to stay put.

But here is where I stop being impressed by the plumbing and start asking what it proves. Scale it against the market it sits inside. India's corporate bond market is roughly ₹53.6 trillion — about 16% of GDP — yet it turns over barely a third of outstanding each year, a turnover ratio of around 0.3 against 60–75% in developed markets. Almost all issuance is private placement, and pension funds, insurers, and mutual funds largely buy and hold to maturity. Retail investors are a rounding error, under 2% of the market.

That ~0.3x number is the whole point, and it is why the pilot's next stage matters far more than the September launch did. SEBI has completed the primary-issuance phase; Stage II, secondary-market trading, is "in the sandbox" and next, with retail access only in later phases. Reuters and others reported that exchanges are expected to stand up a secondary market for these tokenized bonds by December, and the three-month lock-in on the initial issuances naturally expires around then. An allocator deciding whether this model is the scalable template should therefore be watching one question: does Stage II turnover in December rise meaningfully off the ~0.3x baseline, or stay thin because buyers remain buy-and-hold?

I genuinely don't know the answer, and neither does anyone as of this writing — the data point sits in the future. But the two possible outcomes point in opposite directions, which is what makes the metric decisive. If December turnover is meaningfully higher, the case is made: atomic DvP and connected wallets removed a real friction, institutions felt safe enough to trade, and the depository-custodied model delivered the liquidity that open-chain tokenization never could. Then the walled garden genuinely is "the" institutional model.

If December turnover stays thin, the model has failed its own test. A ~0.3x market that remains ~0.3x after being tokenized would suggest the bottleneck was never settlement friction — that it was always a concentration problem, a demand problem, or the simple fact that a handful of AAA insurers and pension funds holding to maturity don't need a liquid secondary market. Faster settlement on an illiquid market is a modest operational win, not a proof that depository-custodied DLT unlocks tokenized-debt scale. It would show up as "efficient plumbing, same buyers."

That is the falsification test, and it is worth stating plainly. The thesis — that legal certainty and custody comfort outweigh lost programmability for institutional tokenized debt — is invalidated by any one of three things: Stage II secondary turnover stagnating near 0.3x; the pilot's scope never extending beyond bonds (SEBI's own roadmap promises expansion to equities, mutual funds, and gold in later stages, but promise is not proof); or institutions and issuers citing composability as the decisive missing piece, which would mean the thing the walled garden gave up actually turned out to matter. Conversely, it is validated by a materially higher turnover ratio in December, and by the extension of the same custodied architecture into the broader financial footprint.

My own read, to be honest, is that India has built the most complete version yet of the "regulated rails" answer to tokenization — and that completeness is exactly what makes the test clean rather than a foregone conclusion. Anyone watching tokenized debt should file this away as a real experiment with a defined pass/fail line, not as evidence the debate is settled. The line is December. If the secondary market for India's tokenized bonds trades like its plain-vanilla cousin, the walled garden will have shown us that custody comfort was never the missing ingredient — and the open ledger's niche status will look less like a problem to be solved and more like a ceiling the system accepts.