2.2 Types of RWAs: A Working Taxonomy
How to classify RWAs by economic exposure rather than wrapper or format, which categories have live institutional products today, and where tokenization changes risk less than common narratives claim.
RWA is not a single asset class. It is a delivery and recordkeeping model applied to asset classes that already have their own legal frameworks, market structures, and risk profiles. A useful taxonomy starts from that observation and sorts tokens by what their holders are actually exposed to. Section 2.1 established the test for whether a token belongs in the RWA category at all. This section sorts the tokens that pass that test into categories that explain how each one behaves, what kind of underwriting it requires, and how mature its market really is.
The temptation is to split assets into physical and digital, or tangible and intangible. Those labels are memorable and weak for analysis. A tokenized Treasury and a tokenized gold bar can use almost identical smart contracts and still demand entirely different questions from an investor. The risk lives in the asset underneath, not in the wrapper around it.
Sort by Economic Exposure, Not by Wrapper
RWA.xyz, one of the main industry data providers, reworked its classification framework around economic exposure for exactly this reason. Wrapper-based labels such as "alternative funds" or "private credit" had grown so broad that they no longer described the risk an investor was taking, and the new framework reorganizes assets by what the holder is exposed to economically 1. The same logic applies to a research taxonomy. Sort by exposure first. Sort by wrapper, chain, or product format second.
The OECD makes a closely related point at the policy level. It defines asset tokenization as the digital representation of real assets on distributed ledgers, or the issuance of traditional asset classes in tokenized form, while explicitly excluding crypto-assets from the definition 2. That distinction is useful here because it anchors the taxonomy to existing asset classes rather than inventing new ones. A tokenized Treasury is a Treasury. A tokenized loan is a loan. A taxonomy that loses that anchor will struggle to say anything useful about risk.
The OECD also flags that the pace and timing of tokenization will vary across asset classes and is likely to follow heterogeneous paths 2. That observation tells the reader to expect uneven adoption: some asset classes will produce live institutional products well before others, and the gap between categories matters more than the headline that "RWAs are growing."
The Categories That Pass the Test
The categories below are listed in rough order of current institutional maturity, not by conceptual appeal or projected market size. The order reflects how much of each category is live today, supported by regulated wrappers, and used by named institutional issuers.
Tokenized Treasuries and Money Market Funds
This is the cleanest entry category and the easiest to underwrite. Treasury bills are short-duration, liquid, standardized, and supported by deep traditional markets, established valuation methods, and decades of regulatory practice. None of those properties come from tokenization. They come from the asset class itself. Placing the asset on a programmable ledger changes the record layer, not the underlying instrument.
BlackRock's BUIDL fund is the working anchor. The fund invests 100% of total assets in cash, U.S. Treasury bills, and repurchase agreements, with a stable $1 per token target value 3. The operating structure is conventional and institutional: BNY Mellon serves as custodian and administrator, and Securitize serves as transfer agent and tokenization platform for subscriptions, redemptions, and distributions 3. The token is a programmable record placed on top of a money market product that already exists inside the U.S. regulatory regime for government money funds.
Two takeaways follow. First, the cleanest tokenized category is the one with the least exotic underlying asset. That is not a coincidence. It is the rule. Second, the operating model is inseparable from the asset class. A custodian, an administrator, a transfer agent, and a placement agent are all named entities with defined responsibilities. Calling a tokenized money market product "DeFi" would miss the point. It is institutional finance with a new record layer.
Credit
Credit is structurally harder than Treasuries because the asset itself is harder. Loans, receivables, structured products, and private fund interests carry credit risk, servicing risk, collateral quality risk, and pricing opacity that Treasuries do not.
Industry trackers have started sorting credit by exposure type rather than by wrapper. RWA.xyz reorganized its credit category into corporate credit, asset-backed credit, diversified credit, and specialty finance, so an investor can tell what kind of borrower or collateral sits behind a given token 1. That is more useful than the catch-all label "private credit," which can describe direct lending to mid-market companies, pools of consumer receivables, trade finance paper, or fund interests in diversified credit strategies, all of which behave differently.
Hamilton Lane illustrates the fund-exposure pattern inside this category. In 2022, Hamilton Lane and Securitize announced tokenized feeder funds giving qualified U.S.-based investors access to funds with exposure to direct equities, private credit, and secondary transactions 4. The token here is a share in a feeder fund. The underlying exposures are illiquid alternatives. The tokenization layer changes how the investor subscribes, redeems, and holds the position, not what they own.
The asset-backed sub-type has its own anchor in the Figure HELOC Token introduced in Section 2.1, an instrument linked to home equity lines of credit and reported on RWA.xyz at a total asset value of about $17.8 billion as of mid-May 2026 5. It is the largest live example of asset-backed credit and the clearest case where the on-chain record sits closer to the authoritative one than a tokenized money market share does, even though the underlying loans still live in conventional servicing systems. The borrower and collateral here (households, residential property) behave nothing like the corporate exposures behind a Hamilton Lane feeder, which is the whole reason the sub-types are worth separating.
Credit tokens may offer higher yields than Treasury products. They also carry borrower default risk, servicer reporting risk, restricted secondary transfer, and valuations that rely on irregular marks rather than continuous price discovery. A reader who treats tokenized credit as "tokenized Treasuries with a higher yield" is reading the category wrong.
Commodities
Commodity-backed tokens, dominated in practice by gold, are the cleanest live example of the dual structure introduced in Section 2.1. The token represents a defined quantity of a physical asset held by a named custodian.
Paxos' PAX Gold (PAXG) is the canonical product. Each PAXG token equals one fine troy ounce of London Good Delivery gold, and PAXG tokens represent fractional ownership of London Good Delivery gold bars held by Paxos Trust on a segregated basis for PAXG holders in LBMA-approved vaults 6. The asset is named. The custody arrangement is named. The legal basis for the holder's claim is named.
The Financial Stability Board generalizes the point without depending only on issuer disclosures. Reference assets that are tokenized may need physical custody, with commodities such as gold given as the example 7. That is the operational price of bringing a hard asset on-chain. Someone has to hold the metal, segregate it from other claims, allow audit and reconciliation, and maintain the link between the token and the bar.
Commodities therefore look simple at the asset layer and demanding at the custody layer. The pattern from Treasuries holds, with the weight moved to a different layer: the asset class is untouched, and what tokenization introduces is a new keeper of the record and a new way for transfers to settle.
Real Estate
Real estate is where the gap between narrative and practice is widest. The FSB confirms that one tokenization project tokenized shares of real estate investment trusts, which is enough to show that REIT-style tokenization exists in production 7. It is not enough to claim that tokenized real estate has reached the maturity of tokenized Treasuries or gold.
Real estate's structural issues outlast the tokenization layer. Property valuations are infrequent and rely on appraisals. Cash flows depend on tenant performance, management quality, maintenance, and insurance. Legal title sits in jurisdiction-specific registries that the smart contract does not control. Secondary transfers are usually restricted by securities law, eligibility rules, or platform policy. None of those constraints relax because the share is recorded on a chain rather than in a transfer agent's database.
The "tokenization unlocks liquidity" claim deserves real skepticism here. A token that represents an interest in a single property has the liquidity of that property minus the friction of the platform, not the liquidity of an exchange-traded security. A token in a real estate fund inherits the redemption rules of the fund, which usually involve queues, gates, or notice periods. The chain layer changes how the position is held. It does not rebuild the underlying market.
Infrastructure
Infrastructure shares many of real estate's properties and adds project-level complexity. Cash flows depend on long-dated contracts, construction milestones, regulatory regimes, and operating performance. Valuations rely on discounted cash flow models updated only periodically. Investors are typically institutional, eligibility is restricted, and secondary markets are thin.
Production examples on public ledgers remain limited, and the most interesting structures today are pilots or private-market issuances accessible only to qualified investors. The category fits the RWA definition. It does not yet behave like a mature market, and a taxonomy that places it alongside tokenized Treasuries on equal footing will mislead the reader.
Art and Collectibles
Art is the noisiest category in mainstream coverage and one of the smallest in serious on-chain RWA data once NFT-native digital art is set aside as a different category. Many art tokenization products are securities sold through marketplace structures, with limited public-chain settlement, restricted secondary transfer, and bespoke legal wrappers.
Two observations help separate the signal from the noise. First, art is illiquid as an asset class. Valuations are sparse, sales are negotiated, and authentication risk is real. Tokenization does not remove any of those properties. Second, the wrapper matters. A token that is a share in a fund that owns art behaves like a fund interest. A token that purports to be direct ownership of an artwork behaves like a fractional title and inherits the legal complexity of that structure. Treat the category as small, fragmented, and structurally illiquid until issuer disclosures and market data say otherwise.
Reading the Taxonomy
The categories above are easier to compare side by side. The columns below surface the underwriting questions, not the marketing surface.
| Category | What the holder is exposed to | Usual legal wrapper | Custody or source-of-truth requirement | Valuation cadence | Liquidity profile | Maturity | Example products |
|---|---|---|---|---|---|---|---|
| Tokenized Treasuries and money market funds | Short-duration government debt and cash equivalents | Regulated fund (e.g., U.S. money market fund) | Conventional custody plus on-chain transfer agency | Daily NAV with stable target value | Subscription and redemption inside fund rules | Production | BlackRock BUIDL, Franklin Templeton BENJI |
| Credit (corporate, asset-backed, diversified, specialty) | Loan or receivable cash flows from underlying borrowers | SPV, securitization vehicle, or feeder fund | Loan servicer plus on-chain transfer agency or issuer contract | Periodic marks, sometimes infrequent | Limited secondary market, usually restricted | Production for some sub-types, emerging for others | Hamilton Lane tokenized feeders, asset-backed credit pools |
| Commodities | Defined quantity of a physical asset (mostly gold) | Trust holding allocated physical asset | Vault custody plus segregated allocation | Continuous market price for gold | Token transferable, redemption rules vary | Production | Paxos PAX Gold |
| Real estate | Property cash flows, equity in a property, or fund interest | SPV or REIT-style structure | Local title registry plus fund or SPV records | Infrequent appraisal-based valuations | Thin secondary market, often restricted | Emerging | REIT-style tokenized vehicles, single-property platforms |
| Infrastructure | Long-dated project cash flows or equity in operating assets | Project SPV, fund, or trust | Project-level books plus fund or trust records | Infrequent model-based valuations | Thin, usually institution-only | Experimental | Pilot issuances, qualified-investor structures |
| Art and collectibles | Share in a fund or fractional title to a specific work | Fund, SPV, or fractional ownership structure | Physical custody plus title and authentication | Sparse, sale-based valuation | Very thin, often platform-restricted | Experimental | Fund-style and platform-specific structures |
The order is honest rather than romantic. Treasuries and money market funds first, credit second, commodities third, then real estate, infrastructure, and art. A taxonomy that puts real estate or art at the top has usually been written from marketing rather than from the data.
Maturity Bands
A taxonomy that only lists asset types implies the categories are at comparable stages. They are not. A three-band view captures the real distribution.
Production. Live institutional products with named issuers, regulated wrappers, and meaningful on-chain activity. Tokenized Treasuries and money market funds sit here, along with the largest gold-backed commodity tokens and some asset-backed credit pools.
Emerging. Live products with credible institutional examples, but the category as a whole still depends on bespoke structures, jurisdiction-specific carve-outs, or limited secondary markets. Tokenized private and structured credit fits here, along with REIT-style real estate tokens.
Experimental. Pilots, single-property issuances, single-fund products, or marketplace-specific structures with little public-chain activity outside the issuer's own platform. Tokenized infrastructure and most art tokenization sit here today.
The bands matter because they tell the reader what to expect when they look at market data. Tokenized Treasuries appear in many trackers because issuers publish disclosures and trades settle on public chains. Tokenized art appears rarely because most of the activity sits on platform-specific systems that do not feed into the same data pipelines. Equal billing in a press release is not equal billing in the market.
Issuer-Sponsored, Custodial, and Synthetic Structures
The taxonomy above sorts by economic exposure. A second cut, sorting by how the token is structured around the underlying asset, is worth noticing because regulators are already using it.
In its January 2026 staff statement on tokenized securities, the SEC distinguishes issuer-sponsored tokenization, where the issuer of the underlying security tokenizes its own instrument, from third-party-sponsored models, which include custodial structures, where a third party tokenizes a security it holds in custody, and synthetic structures, where a token tracks an underlying instrument without granting direct rights in it 8. The reader does not need to memorize each label. What they need is the habit of asking, for any given product, whether the issuer of the underlying instrument is the same entity as the issuer of the token. Issuer-sponsored, custodial, and synthetic models look similar at the wallet level and produce different rights, different redemption mechanics, and different counterparty exposures.
Bridge to the Mechanics
Each category in this taxonomy follows the same high-level tokenization sequence, but the hard part changes from one category to the next. For Treasuries and money market funds, the hard part is fund administration and transfer restrictions. For gold, it is physical custody and allocation. For credit, it is underwriting, servicing, and pool transparency. For real estate, infrastructure, and art, it is valuation, legal title, management, and exit liquidity.
Section 2.3 walks through that sequence as an operational process. Section 2.4 covers the specific machinery, including custody, oracles, and smart contracts, that holds each category together. With the taxonomy in place, those sections can stop explaining what is being tokenized and concentrate on how.
- RWA is not a single asset class. It is a delivery and recordkeeping model applied to asset classes that already have their own market structures, legal frameworks, and risk profiles. A useful taxonomy sorts tokens by economic exposure, not by wrapper, chain, or token format.
- Tokenized Treasuries and money market funds, anchored by products such as BlackRock's BUIDL, are the cleanest live category because the underlying assets are liquid, standardized, and already inside conventional regulatory wrappers.
- Credit is the structurally harder category that often gets lumped together as "private credit." Sub-categories such as corporate credit, asset-backed credit, diversified credit, and specialty finance behave differently and should be underwritten separately.
- Hard assets such as real estate, infrastructure, and art do not become liquid because they are tokenized. Valuation, legal title, management, and exit liquidity remain the binding constraints, and the "tokenization unlocks liquidity" claim deserves skepticism for these categories.
- Maturity bands matter. Production, emerging, and experimental categories sit at different stages of market development, and a taxonomy that treats them as equivalent will mislead the reader about what is actually live today.