Reconciliation — Cross-Module Audit

Read this before quoting any figure. This tab interrogates the six modules against each other: where they disagree, which figures rest on weak sources, where the research is thinnest, and where the evidence revised the working thesis. It is published deliberately, because an evidence base that hides its own weak points is not checkable. Nothing here should be lifted into another document without reading the entry that covers it.

This is a synthesis pass rather than new research, a consistency and defensibility check on what the modules already assert. Where a module's figure rests on T4/T5, that inheritance is flagged; repeating a source does not upgrade it.

1. Contradictions

1.1 Ethereum's share of tokenised RWA value: CONTESTED, do not assert a point figure. Module A says 58–63% (T5); Module D says ~50% (RWA.xyz, T2); Module F says ~65%. Same quantity, three numbers, differing denominators and methods. The best-tier source supports "largest single share, roughly half." Rely on that phrasing; retire the 65% and 58–63% from external use.

1.2 Total tokenised RWA (the denominator). A: ~$33.5bn ex-stablecoins; D: "$17.5–36bn" and "$33.5–38bn"; F: "$31–36bn." The same fast-moving figure on different dates and inclusion rules. Rely on the RWA.xyz ex-stablecoins figure with its exact date, re-pulled at use.

1.3 Broadridge DLR monthly volume. A: $8.0tn (July 2026); D: $7.3tn Jan / $8tn Mar / $7.5tn Jun. A monthly series rather than a conflict, but never quote two months side by side without the month attached. Prefer the Bloomberg/Kaiko feed (T2) over the company release (T4).

1.4 BUIDL AUM: CONTESTED. A: honest "$1.7bn–$2.9bn"; F: "~$2.85bn" as if settled. Rely on the live RWA.xyz entry, quoted as a range if not pulled fresh.

1.5 BUIDL's Ethereum share of its own AUM: genuine conflict. A: "majority on Ethereum mainnet" (Nov 2025); F caveat: "fell to ~40%." These conflict on direction and cannot both be current. Contested; do not assert without a fresh per-chain pull.

1.6 India bank-channel remittance cost: drifting vintage. A/C: 12.66% (Q1 2024); F: notes it rose to 14.55% (Q1 2025) and that it is a global bank average rather than India-specific. Rely on the latest RPW issue, labelled "global bank-channel average."

1.7 e₹ circulation vs. flow: presentation risk. All three modules mix stock in circulation (₹771–1,016 crore) with cumulative transaction flow (hundreds of millions of transactions). Different measures; must be labelled. The FY26 decline in circulation (down 24%) is the key disconfirming fact.

1.8 SEBI DLT covenant system: status safe, scale figures lean on vendor data. The mandate is T1 (the circular); the 4,291-issuer / ~90%-ISIN scale is T4 (Cognizant) plus NSDL IPO docs. Lead with the circular; label the scale numbers.

1.9 SEBI bond pilot timeline: RESOLVED. Announced 26 May 2026 with a "six to nine months" estimate; C/F/G carried an unconfirmed Reuters report of a REC issuance around September 2026. The report was right and the estimate was beaten. REC issued ₹500 crore on 7 September 2026, L&T ₹500 crore and IIFL Finance ₹25 crore on 9 September, and SEBI launched the programme as Demat 2.0 with RBI Governor Malhotra on 10 September (SEBI PR 56/2026, T1). Rely on: launched 10 Sep 2026, ₹1,025 crore, status LIMITED PILOT for primary issuance under SEBI's Regulatory Sandbox; secondary trading and retail access are later phases and have not opened. Never call it IN PRODUCTION.

1.10 "Ethereum leads" (A/D) vs. "Ethereum is the wrong domestic answer" (D/F). Not a contradiction once scoped: A measures the global funds/stablecoin layer; D/F judge the Indian bank-settlement layer. The reconciled framing must travel with any "Ethereum leads" claim: Ethereum leads global asset-tokenisation by value; permissioned ledgers lead bank settlement; India's domestic core-settlement path is CBDC/permissioned, with public Ethereum relevant to GIFT City, cross-border flows, and the domestic fund layer adjudicated at 1.11.

1.11 The domestic fund case: A/C describe a domestic route, D/F's phrasing appears to exclude it. A calls blockchain unit tracking inside an AIF (Cat II) wrapper "the cleanest current route" for the Indian fund analogue and names a tokenised feeder in an AIF wrapper as an alternative to SEBI recognising a DLT register. C's Segment 5 rates that same domestic pathway ANNOUNCED/sandbox-only, with the blocker at VDA classificatory risk and no CBDT carve-out. B rules out mutual-fund units on a public chain domestically, on VDA risk plus non-recognition of a public-chain token as an MF unit under the Depositories Act 1996. D and F scope their negative verdict to confidential bank settlement and put fund distribution on the right-tool side, so the "wrong rail for every domestic use case" reading overstates them. Rely on: the domestic fund case is held by tax classification, and for MF units additionally by register recognition, rather than by confidentiality, finality or sovereignty; GIFT City is the only surface where a fund can be tokenised today, and an AIF wrapper is the nearest domestic route if the CBDT clarifies. Do not present the domestic fund case as available now.

1.12 Module D's privacy negatives against the February 2026 production record: OVERTAKEN, and the finding survives. D's full tier states that Aztec's successor "has no documented regulated-FI production deployment" and that EY's Nightfall has no confirmed regulated production use, and D's 5-minute tier states that "the privacy-focused L2s are not in regulated production." All three were true when written and are no longer. Aztec shipped private smart contracts on mainnet in July 2026; Starknet integrated Nightfall in February 2026; and Archax, an FCA-regulated venue, has distributed tokenised money-market funds from Aberdeen, BlackRock, Fidelity International and State Street under a privacy wrapper on the TEE-based Silent Data L2 since 19 February 2026. Rely on: privacy reached regulated production during 2026 at the fund-distribution layer and has not reached bank core settlement, where Canton still holds the book. Never use the blanket "no privacy L2 is in regulated production" claim, and never use the Archax deployment as evidence about bank settlement. D's 5-minute tier is corrected in place; its full tier carries a dated revision note, per this base's convention that summary tiers reflect the reconciliation pass and full reports stand as researched. Module E carries the detail.

1.13 Module D's "Canton offers no equivalent ladder" and Daml-scarcity argument against Zenith: NARROWING, and the structural case survives. D's 5-minute tier stated that Canton offers no migration ladder comparable to permissioned EVM -> L2 -> public Ethereum, and D leaned on Daml scarcity for India's talent argument. Canton's Zenith layer (announced March 2026, Canton Network blog, T4) runs unmodified Solidity inside Canton with atomic calls into Daml. It is not in production: in July 2026 its external_call() primitive was scheduled for the Canton 3.6 upgrade, with pilots in sandbox and no mainnet date. EthSystems' January 2026 critique (T4, Ethereum-aligned) supplies the other side: consortium governance with no fork, correlated failure when a commit needs every party's node online, and privacy that is trust-based against node operators and governance. Zenith answers none of these. Rely on: Canton's developer-access weakness is narrowing and will largely close for new applications if Zenith ships; its governance, availability and trust-model costs are structural and unaffected. Never say Canton has "no EVM path", and never cite Zenith as live. D's 5-minute tier is corrected in place; the full tier adds Finding 9, a Finding 2 revision note and matrix updates. Both sources are interested parties and are tiered T4.

2. The figure ledger

The full 155-row ledger lives in the Figure Ledger tab, sourced from data/figures.json, filterable by module and tier, with staleness auto-flagged. Audit headline: ~25% of all figures rest on the weakest two tiers (T4/T5), the single most important output of the audit. That share rose from ~22% when Module E was added, because the privacy evidence is unusually dependent on project-reported and crypto-media sourcing. Three are T5-only and flagged UNVERIFIED in-text: the IBBIC 75% LC reduction, the 18% GST-on-services figure, and the GIFT City real-estate deal specifics. The FX convention differs across modules (₹88 in A/B, ₹88.5 in E, ₹95.45 in C), so standardise before any cross-module rupee/USD comparison.

3. Unsupported or weakly-supported assertions

To be sourced or cut before external use: (1) IBBIC "~75%" LC reduction (T5-only); (2) "18% GST on crypto services" (advisory blogs, no CBIC notification); (3) GIFT real-estate deal specifics (CoinGeek-only); (4) "BUIDL Ethereum share fell to ~40%" (no dated per-chain pull, conflicts with A); (5) "~65% of RWA on Ethereum" (not supported by the T2 source); (6) eBG "3–5 days to 2–4 hours" (NeSL vendor claim); (7) Indian conference-attendance constraints (unsourced synthesis, flag as judgement); (8) "RBI was the blocking party" on the discussion paper (anonymous sourcing, present as "reported"); (9) household gold as any addressable figure (price-inflated valuation, not a mobilisable base); (10) "Canton throughput is inflated by repeated repo cycles" (T3 analysis; attribute as analysis, not fact); (11) Digital Asset "$300m at ~$2bn" raise (no clean primary; reconcile against the ~$135m 2025 round); (12) the falsifiable-condition thresholds in Module D (analyst framework, not sourced forecasts); (13) the FIU-IND 2026 AEC and mixer prohibition and travel-rule field list in Module E, which rests on a law-firm summary with the primary FIU-IND circular not located, so the substance is reliable and the exact wording is UNVERIFIED; (14) Aztec's Alpha V5 performance figures (~2.5s proving, <$0.05 fees) and COTI's Privex volume, both project-reported (T4) with no independent verification; (15) the DPDP-versus-immutable-ledger tension, which is analysis rather than sourced law and carries no Indian regulatory or judicial authority; (16) the statement that Demat 2.0's secondary market has not opened, which rests on crypto-media reporting against a T1 phasing document that does not give dates, so treat the phasing as sourced and the current status as needing a fresh check; and (17) the 16 September 2026 Department of Economic Affairs appearance on the VDA framework, T5-sourced, to be confirmed against the Parliamentary Standing Committee's own record before use.

4. The strongest findings

Ranked by how hard a skeptical Indian institutional audience would find them to dismiss, each with its single best source.

  1. India already runs one DLT system in mandatory production in regulated finance (SEBI covenant monitoring). SEBI circular /2022/38, 29 Mar 2022 (T1). Lead with the circular, not the 4,291 scale figure.
  2. SEBI and the RBI jointly launched a permissioned, depository-owned tokenised corporate-bond rail settled atomically in wholesale CBDC. SEBI PR 56/2026, 10 Sep 2026 (T1). ₹1,025 crore from three issuers. The hardest finding in this base to wave away, because it is Indian, recent, primary-sourced, and was done by the two regulators the audience answers to. State the sandbox scope and the unopened secondary leg in the same breath, or it will be taken as an overclaim.
  3. No crypto ban; the Supreme Court struck down the 2018 banking circular. IAMAI v. RBI, 4 Mar 2020 (T1).
  4. The first wave of bank-consortium trade-finance blockchains failed, mostly on governance not technology. TradeLens wind-down, Q1 2023 (T3).
  5. The highest-volume tokenised-settlement system runs on a permissioned ledger, not public Ethereum. Broadridge DLR via Bloomberg/Kaiko (T2).
  6. Confidentiality is the on-record reason JPMorgan, Goldman, DTCC, HSBC chose Canton over Ethereum. Yuval Rooz, The Defiant, Dec 2025 (T3) plus DTCC SEC No-Action Letter (T1).
  7. India's near-term pilots settle in wholesale CBDC, not stablecoins. RBI Annual Report FY26, 29 May 2026 (T1).
  8. Indian infrastructure adoption historically required compulsion or subsidy; demat scaled only under mandate. SEBI Annual Report 1998-99 (T1).
  9. Account Aggregator shows a consent network without a mandate produces enablement without usage, at <10% conversion after five years. Sahamati dashboard, Mar 2026 (T2).
  10. UPI's "free" model isn't self-financing; subsidy covers ~11% of industry cost. Parliamentary Standing Committee on Finance, 32nd Report, 12 Mar 2026 (T1).
  11. The largest 2025 crypto loss ($1.5bn, Bybit) was key-management/supply-chain, not an Ethereum protocol failure; DeFi protocol losses fell 74%. Chainalysis 2026 Crypto Crime Report (T2) plus Immunefi 2026 (T2).

5. The weakest points

Where the research is thinnest and where an informed opponent attacks first.

  1. Every vendor volume claim is company-reported and unaudited (Kinexys, Citi, Broadridge, HSBC). Fix: a Bloomberg/filing confirmation for at least Kinexys and Broadridge; label as company disclosure until then. The Broadridge Bloomberg/Kaiko feed (T2) is the one upgrade available now.
  2. Ethereum's RWA share is quoted at 50/58–63/65% across modules. Fix: commit to one measure (RWA.xyz distributed value), one date, "largest single share, roughly half"; retire the others from external use.
  3. No measured Indian cost-saving from any live pilot. The report sizes markets well but has no rupee-saving figure from any pilot; eBG/IBBIC savings are vendor/T5. Fix: RBI/SEBI pilot post-mortem data or an NSDL/CDSL operational metric; until then concede openly.
  4. The conference/Devcon thesis rests on absence of evidence and unsourced attendance-constraint synthesis. Fix: surface one sourced conference-to-pilot causal case, or hold the honest "convening, not procurement" line and do not overclaim.
  5. The legal-cure path is long, unowned, and partly speculative (PSS Act finality, Depositories Act register, CBDT VDA carve-out, none with a named sponsor). Fix: separate the administratively-achievable (RBI can designate under the PSS Act) from the legislative (SEBI/Parliament/CBDT); do not overstate near-term feasibility.
  6. The privacy evidence rests largely on one consortium report whose members are the vendors it rates. The EEA Privacy Working Group's April 2026 survey is the best available independent classification, applies a strict standard ("No public evidence, no production claim"), and is produced by a body the rated vendors belong to. Fix: corroborate at least the general-availability and early-production classifications against a customer disclosure or a filing; label the source whenever the status table is quoted.
  7. The one regulated privacy deployment runs on a hardware trust model rather than a cryptographic one. Silent Data is TEE-based, so its guarantee depends on manufacturer attestation rather than a proof, a materially weaker assumption than the Canton comparison implies and than most retellings of the Archax deployment convey. Fix: state the trust model whenever the deployment is cited, and do not let it stand in as evidence that zero-knowledge privacy is production-ready.

6. What changed the thesis

Working position: Ethereum as a neutral settlement/interoperability layer beneath India's existing infrastructure, complementing not replacing it; the case is about infrastructure, not holding an asset.

Supports it. The infrastructure-not-asset half is vindicated cleanly. India's regulators mandate and pilot DLT infrastructure while remaining hostile to crypto as an asset, so the asset/infrastructure line is the actual line regulators draw, and the single most important thing to get right. "Complementing not replacing" holds for the one production case: SEBI's system sits inside the depository perimeter, layered on rather than substituted, the demat pattern.

Complicates it. "Neutral settlement layer" overreaches for the domestic market. For core bank settlement, the market chose permissioned ledgers and CBDC settlement over a neutral public chain, globally and in India's own pilots, for reasons of confidentiality. "Interoperability layer" is aspirational: no Indian production case uses Ethereum for interoperability; that value sits offshore behind GIFT City and the finality gap. SWIFT's decision sharpens the same point in Ethereum's favour and against the public-chain reading at once: its shared ledger is built on Linea, an Ethereum layer-2 stack, and went to seventeen named banks in July 2026 as permissioned infrastructure. Ethereum's execution technology and talent pool won a decision at the centre of interbank messaging while public Ethereum settlement lost it, which is the strongest evidence yet for the three-way framing and against treating this as a public-versus-permissioned argument.

Contradicts it. The premise that Ethereum specifically is the substrate does not survive the production record. For every domestic core-settlement use case examined, the honest answer is that a permissioned ledger or the CBDC is the right rail and public Ethereum is the wrong one on confidentiality, finality, and sovereignty. The domestic fund case turns on different grounds: C's Segment 5 puts the blocker at VDA classificatory risk with no CBDT carve-out, and B records that no public-chain token is recognised as a mutual-fund unit, so tax classification and register recognition hold that case rather than confidentiality or finality. Public Ethereum's defensible domestic role is GIFT City fund tokenisation and genuinely cross-border flows, with blockchain-based unit tracking inside an AIF wrapper the nearest domestic route once the tax question is settled.

Stated plainly: the evidence does not support Ethereum as the general-purpose settlement layer beneath Indian finance. It supports Ethereum as one venue: leading globally at the asset-tokenisation layer, marginal at the Indian domestic-settlement layer, relevant to India mainly through GIFT City, cross-border flows, and regulated fund wrappers. Revise the working position to: "distributed-ledger settlement as an emerging layer, on which India will use sovereign/permissioned rails for domestic core settlement, public Ethereum offshore, and, once the tax classification is settled, public-chain unit tracking inside already-regulated domestic fund wrappers." The infrastructure-not-asset half is confirmed; the Ethereum-as-neutral-domestic-layer half is not.

7. Form-factor map

Which findings anchor which output. Numbers refer to the strongest findings above; ledger IDs refer to figures.json.

  • One-page conference leave-behind: #1 (SEBI DLT in production); ledger #54+#56 (₹53.6L cr bonds / ₹7,645 cr/day, illiquidity in one line); ledger #60 (US$129bn remittances); #4 (Broadridge on permissioned); the §1.10 bifurcation as the one-line thesis.
  • Fifteen-slide bank technology committee deck: #1, #4, #5 (production reality + public/permissioned split); ledger #85+#86 (Canton $344.8bn represented vs $961k public TVL); ledger #54/#56 + Module C gap analysis; #3 (graveyard) as the risk slide; weakest points #1 and #3 as the "what we don't yet know" slide.
  • Policy note for a regulator's technical staff: #6 (RBI FY26 AR); the finality gap (PSS Act 2007) and register gap (Depositories Act 1996); ledger #111 (FATF R.15 65% non/partial); comparative mechanisms (UK DSS, EU DLT Pilot, Switzerland ledger-based security); the two-way sovereignty analysis.
  • Op-ed for the Indian financial press: #1 (India already runs a production blockchain, the hook); #7 (demat precedent) + #9 (UPI subsidy ~11%); ledger #40 (e₹ circulation fell 24%, the honest disconfirming detail); the §1.10 bifurcation as thesis.
  • Briefing for a non-beat journalist: #2 (no crypto ban) to clear the biggest misconception; #1 (SEBI already mandates a blockchain) as the surprise; ledger #75 (PNB–Nirav Modi ₹14,356 cr) as the concrete story, labelled, since its sourcing is T5/T3; the asset-vs-infrastructure distinction as the framing they must not get wrong.
  • Session abstract for a developer conference's institutional track: #4 + #5 (bifurcation, Canton's confidentiality win, which signals honesty); ledger #90/#91 (India developer depth) as the "why here" hook; the Module G format precedent (closed roundtable, regulator-anchored); weakest points #4 (honest ceiling: convening not procurement) to set expectations.

Fix these before anything external ships: (1) standardise the Ethereum-RWA-share claim to one measure/date and retire the 65% and 58–63% variants; (2) attach a tier label to every vendor volume figure and never chart on a T4/T5 number alone. What would most strengthen the whole body: a Bloomberg/filing confirmation of one bank volume, and any measured Indian pilot cost/benefit figure, currently the biggest evidentiary hole.