TL;DR
This playbook is for the product lead, founder, or bank programme owner picking between tokenisation issuance platforms (or weighing the build-vs-buy alternative) for a single named programme. The decision is not generic. Tokenised fund issuance, tokenised security issuance, and tokenised deposit issuance are three different decisions with three different shortlists and three different gating criteria. A platform that runs BlackRock's BUIDL beautifully is not the same shape as a platform that runs JPMorgan's BDA, and neither is the right answer for a Reg S tokenised credit fund into APAC accredited investors. Pick the wrapper before you pick the platform, and expect the platform shortlist to fall out of the wrapper rather than the other way around.
Decision frame
The first question is build versus buy. In-house build makes structural sense in three configurations. A GSIB-scale balance sheet running tokenised deposits as a bank-money product (the Kinexys pattern, the Progmat consortium pattern, the Partior multi-bank pattern). A multi-asset-class roadmap that will run tokenised funds, tokenised securities, and tokenised collateral over the same operating perimeter and where the cost of integration with three vendors exceeds the cost of one platform team. A distribution franchise where the platform itself becomes a product (the Securitize case, the Ondo case post-Oasis Pro). Outside those three, in-house build tends to be a strategic mistake at single-product scale, and the platform vendors are the right counterparty.
Buying makes sense when the programme is a single-product line in a single jurisdiction, when speed-to-market matters more than long-term unit economics, and when the asset manager's distribution franchise does not depend on the platform being proprietary. The asset-class question follows. Tokenised fund interests sit on a different platform set (Securitize, Ondo, Centrifuge, Tokeny) than tokenised securities (the same set plus Provenance, plus exchange-led platforms like the NYSE-Securitize partnership) than tokenised deposits (bank-only: Kinexys, Partior, Fnality, Progmat). Each carries its own regulatory wrapper and its own platform sweet spot.
Evaluation dimensions
Registrations the platform holds
The load-bearing question. A platform without the relevant registrations cannot host a genuinely-tokenised instrument where the on-chain entry is the legally operative record. For a US fund or security, the SEC-registered transfer agent registration is the gate (see Chapter I, Part 3 on legal control and the transfer-agent theme). Without it, the platform can run a representation of an off-chain register, but the off-chain register controls in any disputed case, which defeats the structural purpose of tokenising. Securitize and (post-Oasis Pro) Ondo hold the SEC transfer-agent stack plus broker-dealer plus ATS (alternative trading system); most US tokenisation tech vendors do not.
For non-US perimeters the equivalent gating registrations are the Major Payment Institution licence in Singapore for any payment-rail tokenisation, FINMA registration for Swiss DLT-securities, BaFin for German crypto-securities under the eWpG, and the relevant SFC licence in Hong Kong for any tokenised collective investment scheme. Money transmitter or money services business licences cover the US payments side. Ask the platform for the entity, the registration number, and the regulator. A vendor that hand-waves on this is not a serious counterparty for a genuinely-tokenised programme.
Chain coverage
Multi-chain has become the operating default for institutional tokenised funds. BUIDL is on nine chains. FOBXX, Apollo's ACRED, OUSG all operate across Ethereum mainnet plus a meaningful subset of L2s plus Solana. The question is whether the platform supports the chains where your distribution channel actually settles, not whether it claims the longest list. If your buyer set is concentrated on Ethereum mainnet and Base for institutional flow, then Ethereum mainnet and Base are the chains that matter. If you need Solana for a specific stablecoin-denominated subscription rail, then Solana matters too. If the institutional cash-rail counterparty is on Canton or a permissioned EVM ledger like the Kinexys rail, the platform needs to bridge there cleanly or the integration is off-chain and brittle.
A specific question worth asking: how does the platform handle whitelist consistency across chains. A holder on Ethereum and a holder on Solana need to be the same legal counterparty under the relevant transfer-agent recordkeeping rules even though the chain identities differ. Platforms that have not solved cross-chain whitelist administration cannot support the multi-chain pattern at institutional grade.
Asset-class fit
Platforms specialise. Securitize's centre of gravity is institutional MMFs and private fund interests under the 3(c)(7) wrapper, with the BUIDL reference architecture as the canonical example (see BUIDL reference architecture). Centrifuge's centre of gravity is the SPV-plus-tokenisation pattern on public chains, with the Janus Henderson AAA CLO and the Anemoy MMF as the worked examples and Sky's Grove allocator as the major DeFi-side counterparty (see Centrifuge V3 + Aave Horizon). Ondo issues its own products (OUSG, USDY, Global Markets equities) and is now a platform layer through the Oasis Pro stack. Tokenised deposits are not a platform-vendor decision: they require the issuing entity to be a bank, and the platform is either in-house (Kinexys, Progmat) or a consortium rail (Partior, Fnality).
For tokenised securities (bonds, equities, structured notes), the platform set overlaps with the fund set but adds exchange-led infrastructure (the NYSE-Securitize partnership, SDX in Switzerland, the Korean and Japanese megabank consortia). For tokenised private credit specifically, Centrifuge has the most public-chain-native production references. Treat asset-class fit as a pass/fail filter rather than a soft preference.
Distribution rails
Ask which broker-dealers, wealth platforms, private banks, and ATS venues the platform plugs into out of the box, and quantify how much of the distribution work the platform is doing versus what your team has to do. Securitize's distribution lane runs through Securitize Markets (the in-house broker-dealer and ATS) plus the institutional buyer base anchored by BUIDL's holder set. Ondo's distribution runs through DeFi-native channels for OUSG and USDY, plus the MetaMask integration for Global Markets, plus the Oasis Pro broker-dealer infrastructure. Centrifuge's distribution runs through public-chain DeFi venues, including the Aave Horizon institutional-collateral lane that launched with JAAA, JTRSY, and BUIDL as launch collateral.
The question is which of these matches your buyer base. A private-bank-distributed product needs different rails than a DeFi-native product. A platform that brings the right distribution lane is doing meaningfully more work than a platform that just issues the contract.
Secondary trading venue partnerships
Most tokenised funds do not have meaningful secondary trading. Subscriptions and redemptions clear at NAV against the issuer (or its manager contract), and the on-chain liquidity primitive is either absent or a thin atomic-swap construction (the Circle USDC swap adjacent to BUIDL is the cleanest published example). For tokenised securities the secondary picture matters more, and the question becomes which ATS the platform's tokens can trade on, and whether that venue has institutional liquidity or is primary-market-only. Securitize Markets is one such venue; Oasis Pro is another. Aave Horizon is not a secondary venue but functions as a borrowing-against venue, which can substitute for some of the liquidity function in DeFi-native distribution.
Jurisdictional reach
US perimeter only is a viable shortlist for some products. Cross-border qualified-investor distribution (Reg S, MiFID II, AIFMD, MAS SFA Section 274 / 275, HK SFC professional investor regime) requires the platform to either hold the equivalent registrations in the destination jurisdictions or work with local counterparties that do. Ondo's Global Markets is the most credible recent example of a tokenised-equities product reaching non-US qualified investors at scale through MetaMask, with structural reach into APAC, EMEA, LATAM, and Africa. Securitize's reach is primarily inside the US perimeter, which is a feature for a US-anchored programme and a constraint for a programme that needs APAC distribution.
Fee model
Platforms charge across at least four cost dimensions. Per-issuance setup fees (one-time, sometimes substantial). AUM-based fees on the live mandate (predictable scaling). Per-investor fees (often hidden, can compound rapidly with retail-style growth). Per-transaction fees on subscriptions, redemptions, and transfers (the cost dimension that bites a high-velocity product). Build the cost stack against your AUM curve and your investor-count curve. A fee model that looks reasonable at USD 100m AUM and 50 investors can become unreasonable at USD 1bn AUM and 5,000 investors if the per-investor fee dominates the AUM-based fee.
Vendor lock-in posture
Three questions. Can you take your token contract off the platform if you choose to leave, or is the contract structurally tied to the vendor's infrastructure. Is the smart-contract source under your control and your auditor's, or under the vendor's. What is the data export path for the holder register, the transaction history, and the KYC evidence base. Vendor lock on the token contract is the gating risk: if you cannot redeploy the contract elsewhere, you do not actually own the instrument. Securitize's Vault Registrar EIP work is positioned partly as a way to reduce this lock at standards level, but the structural question for any platform is whether the contract you launch with is portable.
Operational maturity
Production track record on your specific asset class. References from comparable customers willing to talk on the record (or close to it). Incident response history, including any chain-level or contract-level incidents and how they were handled. Hosting equity issuances does not translate to hosting bond issuances. Private-credit references do not translate to MMF references. Treat the reference call as the most important diligence step, ahead of the platform demo.
Worked example
An APAC asset manager scoping a tokenised credit fund product targeting both US qualified purchasers and APAC accredited investors evaluates Securitize and Centrifuge as the two serious shortlist platforms.
Securitize scores strongly on registrations (SEC transfer agent, broker-dealer, ATS bundle in one counterparty), on US-perimeter operational maturity (the BUIDL reference architecture is the canonical example of how to run a 3(c)(7) wrapper end-to-end), and on chain coverage (BUIDL's nine-chain footprint with whitelist consistency is operationally proven). It scores less well on public-chain DeFi composability (the wallet-allowlist gating forecloses arbitrary DeFi distribution and means BUIDL's path into Aave Horizon required a dedicated institutional-collateral lane). It scores less well on APAC distribution depth: the platform's centre of gravity is the US perimeter, and APAC accredited-investor distribution is feasible but not the platform's strongest lane.
Centrifuge scores strongly on public-chain native distribution (V3's nine-ledger EVM migration completed in April 2026, with the Aave Horizon institutional-collateral lane and the Sky / Grove allocator integration as the worked DeFi-distribution channels). It scores strongly on tokenised credit specifically: the Janus Henderson AAA CLO crossed USD 1bn TVL, which is the public-chain credit-product credibility threshold. It scores less well on US-perimeter regulatory infrastructure: the platform is a public-chain-protocol perimeter rather than an SEC transfer-agent perimeter, which means the US qualified-purchaser distribution requires a separate transfer-agent appointment or a Reg D/Reg S structure that does not need the on-chain entry to be the operative SEC share register.
The recommendation lands differently depending on the distribution thesis. If DeFi-native distribution is core (the asset manager wants the product to be borrowable-against on Aave Horizon, available to DeFi-native stablecoin reserve allocators like Sky / Grove, and natively present on the chains where institutional DeFi liquidity concentrates), Centrifuge is the cleaner fit, with a separate transfer-agent appointment for the US perimeter. If distribution is private-bank-only with no DeFi requirement, Securitize is the cleaner fit because the SEC transfer-agent integration removes the single largest US-perimeter friction and the platform's mandate set is concentrated in exactly that buyer profile. The asset manager should not assume a single answer.
Red flags
- Missing transfer-agent registration when the legal model requires the on-chain entry to be the operative record. The most common gating fail and the one that tends to surface late in diligence.
- Single-chain commitment when the distribution channel requires multi-chain. A platform whose architecture assumes Ethereum-only or single-permissioned-ledger-only cannot support the multi-chain pattern that has become the institutional default.
- Fee structures that scale unpredictably as AUM grows. Per-investor fees that quietly outpace AUM-based fees as the holder count expands. Per-transaction fees that bite a high-velocity product.
- Vendor lock on token contracts. If you cannot redeploy the contract elsewhere, the platform owns your instrument.
- No production references on the relevant asset class. Hosting equity issuances does not translate to hosting bond issuances. Private-credit references do not translate to MMF references. The reference set has to match the asset-class shortlist.
- Marketing materials that conflate "tokenisation platform" with "DeFi protocol". They are different products with different regulatory perimeters, different operating disciplines, and different counterparty risks. A vendor that markets itself across both categories without distinguishing them is signalling it does not understand its own positioning.