RWA Tokenisation's Real Bottleneck Isn't the Blockchain

The numbers look like the thesis is already proven. Tokenised U.S. Treasuries crossed $11 billion in March 2026, up from $7.4 billion at mid-2025. BlackRock's BUIDL fund holds $2.9 billion. Circle's USYC overtook it briefly at $2.2 billion. Goldman Sachs is using tokenised Treasuries as collateral in derivatives transactions, not as a proof of concept but as infrastructure. The firm announced plans for 24/7 tokenised Treasury and money market trading. Total tokenised real-world assets on-chain sit around $25 billion.

The headline version of this story is that real-world assets are going on-chain and TradFi has arrived. That version is not wrong. But it obscures a pattern that matters more: the assets that are easiest to tokenise are the ones that needed tokenisation least, and the assets that would benefit most are barely moving.

The evidence#

The growth in tokenised RWAs is almost entirely concentrated in two asset classes: government securities and private credit instruments. These have clear legal structures, standardised documentation, well-understood custody chains, and, critically, issuers large enough to absorb compliance costs.

Treasuries are the easiest case. The underlying asset is a U.S. government obligation. There is no valuation dispute. Custody is simple. Redemption is predictable. The token is essentially a wrapper around an already-liquid, already-trusted instrument. What tokenisation adds is programmability, composability with DeFi protocols, and 24/7 settlement. Those are real benefits. But they are efficiency gains on an asset that already works.

The pattern holds for institutional private credit. Firms like Apollo, Hamilton Lane, and KKR have tokenised fund shares, giving accredited investors fractional access and faster settlement. But the underlying legal structure is still a traditional fund with a traditional GP, traditional custody, and traditional redemption terms. The token sits on top. It does not replace the plumbing underneath.

Now look at what is not scaling:

• Tokenised real estate remains a fraction of total RWA value on-chain. Grant Cardone announced plans to tokenise a $5 billion real estate portfolio in February 2026, but announcements outnumber functioning secondary markets by a wide margin. Platforms offering real estate tokens report thin liquidity, redemption queue problems during downturns, and regulatory friction around lawful secondary trading.

• Tokenised equities remain largely experimental. The legal question of whether a token constitutes a security, a receipt, or a contractual claim varies by jurisdiction and has not been settled in most of them.

• Commodities tokenisation exists but is concentrated in gold-backed tokens (Paxos, Tether Gold) that function more like ETF alternatives than true on-chain commodity markets.

The disproportion is telling. The largest share of tokenised value sits in the asset classes that already had the most efficient markets. The asset classes where tokenisation could free up the most value (illiquid real estate, private equity secondaries, infrastructure debt) are the ones where the off-chain plumbing is hardest to replicate.

Why this pattern holds#

The bottleneck is not minting a token. Any competent team can deploy an ERC-20 that represents a claim on an asset. The bottleneck is everything that makes that claim enforceable, redeemable, and tradeable.

Legal enforceability. In most jurisdictions, a token does not confer direct ownership of the underlying asset. It represents a contractual claim, typically mediated through an SPV, a trust, or a custodial agreement. If the issuer or custodian fails, token holders may find themselves as unsecured creditors rather than asset owners. The legal architecture that prevents this (segregated SPVs, regulated custodians, third-party audits, proof-of-reserves) is expensive, bespoke, and jurisdiction-dependent. It does not scale the way software does.

Custody and servicing. A tokenised Treasury needs a custodian to hold the bond and an oracle or attestation mechanism to confirm the reserve. That is manageable. A tokenised commercial property needs someone to collect rent, manage maintenance, handle tenant disputes, deal with liens, manage insurance, and report on all of it in a way that maps to on-chain token holder rights. The lifecycle complexity is orders of magnitude higher.

Redemption design. Treasuries can be redeemed at par on predictable schedules. Real estate tokens cannot. When property values decline and redemption requests spike simultaneously, platforms face a liquidity mismatch that no smart contract can resolve, because the underlying asset is a building, not a balance. The redemption queue problems that platforms report during downturns are not bugs; they are the structural reality of tokenising illiquid assets without solving for liquidity at the design layer.

Secondary market infrastructure. For a token to be liquid, someone needs to make a market in it. Tokenised Treasuries have market makers because the underlying is a $25 trillion market with transparent pricing. Tokenised shares in a single commercial property in Manchester do not. The promise of fractional ownership creating liquidity only works if there are enough buyers on the other side. For most niche real-world assets, there are not.

Regulatory fragmentation. MiCA provides a framework in the EU. The GENIUS Act provides one for stablecoins in the U.S. But there is no equivalent framework for tokenised securities, real estate, or private credit that is internationally harmonised. Cross-border trading of tokenised RWAs remains a compliance exercise, not a product feature.

What breaks it#

This pattern breaks if three things happen. There are early signs of each.

Standardisation of legal wrappers. Trust companies are emerging as the backbone of RWA tokenisation, providing the off-chain legal layer that makes on-chain claims enforceable. If the industry converges on a small number of standardised trust structures (the way securitisation converged on standardised SPV templates in the 1990s), the cost of bringing new asset classes on-chain drops significantly.

Institutional secondary markets. Goldman Sachs spinning out its tokenisation platform (GS DAP) into shared infrastructure is a signal. If tokenised asset trading moves from proprietary platforms to shared institutional venues (with standardised settlement, clearing, and compliance), the liquidity problem starts to look different. The infrastructure does not exist yet at scale, but the direction is visible.

Regulatory convergence. If major jurisdictions agree on what a tokenised security is, who can custody it, and how secondary trading is governed, cross-border liquidity becomes possible. MiCA is a start. The GENIUS Act is a start. But neither covers the full scope of tokenised RWAs. The gap is where the friction lives.

None of these are guaranteed. Standardisation takes years. Institutional venues need critical mass. Regulatory convergence across the U.S., EU, UK, Singapore, and the Gulf states is a diplomatic exercise as much as a legal one. The timeline is measured in policy cycles, not product sprints.

What readers should take from this#

The RWA tokenisation trend is real, but the market is pricing in the destination without accounting for the journey. The $11 billion in tokenised Treasuries proves that blockchain rails can handle regulated, standardised assets. It does not prove that those same rails can handle the messier, more valuable asset classes where tokenisation would actually be transformative.

The projects and protocols that will win this market are the ones solving the unglamorous problems: legal enforceability across jurisdictions, custody for complex asset lifecycles, redemption design that survives stress, and secondary market liquidity for assets that have never had it. Not the ones with the best token minting infrastructure.

Watch what Goldman, BlackRock, and the trust companies build next. The token is the easy part. The legal and operational scaffolding underneath it is where the value (and the bottleneck) actually sits.