Centrifuge does not sell one particular bond or operate a single fund. It provides infrastructure for creating fund shares, recording their accounts and distributing them across several blockchains. Its documentation names Apollo, Janus Henderson and S&P Dow Jones Indices among the institutions whose strategies it supports. It also says more than $2 billion of real-world assets have been tokenized through the platform.
That number is impressive, but it cannot evaluate the project on its own. At 04:52 UTC on 26 August 2026, DeFiLlama tracked $1.64 billion of value in Centrifuge. The two measurements are not necessarily contradictory: “assets tokenized” may cover a different scope or period from the value currently tracked by an aggregator. Yet the public documentation reviewed for this portrait does not reconcile them precisely. That difference is a useful starting point for understanding Centrifuge. The protocol makes many operations visible, but it does not turn every legal and financial reality into a single onchain number.
What Centrifuge actually builds
A manager creates a pool in Centrifuge, the technical envelope of a product. It can define several share classes, the assets accepted for subscriptions, transfer rules, valuation frequency and redemption conditions. Each class becomes an ERC-20 token. Vaults compatible with standards such as ERC-7540 handle asynchronous requests: an investor deposits an asset, the manager accepts the request, and shares are then issued. Redemption follows the reverse route and can therefore take longer than an instant token swap.
This stack addresses a real problem. A manager distributing the same fund on Ethereum, Base or Avalanche should not have to maintain nine incompatible registers and manually reconcile nine balances. Centrifuge uses a hub-and-spoke architecture. The hub keeps consolidated accounts, prices, holdings and requests. Spoke chains host tokens, vaults and investor interactions. When a share moves from one chain to another, it is burned on the first and minted on the second, in principle in the same quantity.
The result is more specific than a token that merely represents an asset. It is a fund-operations layer: approved-investor lists, double-entry accounting, net asset value, subscription and redemption orders, transfer restrictions and cross-chain messages. As with the [Indian tokenized-bond pilot examined yesterday](/en/articles/inde-pilote-obligation-tokenisee-deux-portefeuilles), making the register faster does not remove access, settlement or control procedures.
Nine networks, but highly concentrated activity
Centrifuge’s documentation says version 3.1 is deployed on nine networks. Independent DeFiLlama data confirms non-zero value on nine chains, but above all reveals concentration. Ethereum accounted for $1.278 billion, or 77.9% of the tracked $1.64 billion. Avalanche added $261.6 million, or 16.0%. Base held $57.9 million. The six remaining networks together represented about $42.2 million, less than 2.6% of the total.
That distribution qualifies the word “multichain”. The technical ability to distribute a fund across nine networks exists; liquidity is not spread evenly among them. This matters to an investor assessing whether entry or exit is practical on a particular chain, and to a manager as well. Every deployment adds messages, gas costs, adapters and controls without guaranteeing local demand.
Marc Norat’s own analysis is to separate three measurements that are often blended together. The tokenized amount describes what the platform says it has issued or served. Tracked value measures what a third party can attribute to the protocol at a given moment. Distribution by chain shows where that value actually sits. Centrifuge clearly passes the test of observable use, but the $2 billion figure should not be read as $2 billion of immediately accessible liquidity on each of nine networks.
The register is onchain; the claim is not fully onchain
The protocol can automate bookkeeping and make share movements visible. Code alone cannot guarantee that the stated asset exists, that a custodian protects it, that its reported value is correct, or that the legal vehicle gives every holder the expected rights. Those responsibilities remain divided among the manager, issuer, administrators, custodians, oracles and product contracts.
Centrifuge’s customizability makes this boundary especially important. Its documentation says pool managers can configure holdings, share classes, prices and synchronization across chains. Transfer controls can enforce allowlists or block addresses after an offchain KYC check. This flexibility helps a product comply with financial law. It also means two products created with the same protocol can have radically different risks, fees, delays and legal rights.
There is therefore no universal “Centrifuge” yield or safety profile. The correct unit of analysis remains the fund: offering document, issuing entity, assets, custodian, valuation method, redemption calendar and manager powers. Blockchain improves the traceability of part of the process; it does not replace that due diligence.
Faster governance, with less direct decentralization
Since proposal CP171 was approved on 3 November 2025, active DAO governance has been paused. The Centrifuge Network Foundation oversees the protocol and treasury, while Centrifuge Labs executes development, partnerships and operations. CFG holders retain a mechanism to reactivate the DAO, involving a discussion, a proposal, a seven-day Snapshot vote and a quorum of 4 million CFG.
The arrangement reduces coordination delays but concentrates execution in a board and a service company. CP171 projected $15 million of revenue for 2026; that was a forecast published in October 2025, not an audited result. The protocol supports customized fee mechanisms, but there is no universal schedule from which revenue can simply be derived from tokenized value. The public reports promised by the foundation will therefore be needed to connect adoption, revenue, treasury spending and any value captured by CFG.
The token introduces another distinction. In June 2026, Centrifuge reported a total supply of 697.2 million CFG, annual inflation of 3% paid to the treasury, and 54.6% of supply considered released. Holding CFG is neither ownership of fund assets nor a claim on fund income. It is an ecosystem and governance asset whose economic link to activity depends on foundation decisions.
Audits reduce software risk; they do not close it
The public protocol repository lists more than twenty security reviews. A Sherlock review conducted in April 2026 on the Onchain Portfolio Manager found one medium vulnerability that could falsify a loss measure and block some operations. The report marks it as resolved. A separate Burra Security review identified two medium issues, also fixed, plus a design limitation: deposits occurring while a strategy runs can change the aggregate balances used to measure slippage and mask a loss. The team described a complete fix as impractical under that design.
These findings are more indicative of an active control process than proof of a defective system. They nevertheless expose three limits. An audit covers one scope and one version. The public repository is canonical for integrations, but the project says active development takes place in a private repository before selected changes are synchronized. Finally, no smart-contract audit certifies offchain assets, fund managers or legal documents.
The cross-chain design widens the operational perimeter too. A single product may depend on its hub, spoke contracts, price updates and several messaging adapters. The architecture can aggregate multiple providers and retry messages, but more routes also mean more states to monitor. A token displayed on one chain does not prove that every accounting update, redemption order or legal record has reached the same final state elsewhere.
What to verify next
Centrifuge has a working product, verifiable deployments and significant value. It is neither a mock-up nor a bare tokenization promise. Its task is no longer to prove that fund shares can circulate onchain. It must show that this circulation sustainably improves distribution, transparency and operating costs without moving risk into less visible corners.
Four pieces of evidence would strengthen the portrait: a definition reconciling the more than $2 billion tokenized with current tracked value; public fund-level data on subscriptions, redemptions, delays and fees; the revenue and treasury reports promised under the new governance structure; and continued disclosure of the limitations identified in audits of automated managers. Until then, Centrifuge is best understood as RWA infrastructure with observable use, while every token remains a map to a legal product whose underlying terrain still requires inspection.