Top RWA Protocols Compared: Features, TVL & Token Analysis
Pick by need: Ondo for Treasury yield, Maple for institutional credit, Centrifuge for asset origination, Pendle for yield trading, Goldfinch for emerging-market lending - TVL, token data, and risk compared.
Introduction
$38.32B in real-world assets sits tokenized on-chain in August 2026, and every governance token belonging to the protocols that built that market trades between 78% and 99% below its all-time high (rwa.xyz and CoinPaprika, 2026-08-22). The gap is not the only thing that changed since these protocols were first ranked side by side. One of the six, Goldfinch, has voted itself into wind-down, another has lost 91% of its deposits, and a third halted its own blockchain for thirty hours after an exploit. This comparison covers Ondo Finance, Maple Finance, Centrifuge, Goldfinch, Pendle, and MANTRA on what each actually holds today, where the yield comes from, what the token does or does not entitle you to, and which of them are still worth comparing at all.
Key Takeaways
- Ondo Finance and Maple Finance are the two protocols that grew. Ondo holds about $3.49B in total value locked and Maple reports $4.65B in assets under management, up 81% year over year, against a DeFi lending market that shrank.
- Goldfinch voted to wind down in June 2026. Roughly $100M was originated over the protocol's life, about $56M of borrower capital remains outstanding against $1.63M of TVL, and the stated recovery horizon is "two or more years."
- Pendle's TVL fell from a $13.4B peak in September 2025 to about $1.17B, a 91% decline, as the Ethena incentives that had inflated it unwound. It is still the largest yield-trading protocol; it is no longer the largest anything else.
- The whole RWA yield range has compressed into roughly 3.2% to 5.2%. The old picture of 3–5% Treasuries against 8–15% private credit no longer describes what these protocols actually pay.
- Governance tokens did not follow the assets. PENDLE, ONDO, CFG, MPL, and GFI sit 78%, 83%, 94%, 99%, and 99% below their highs while the sector they serve more than doubled.
What Are RWA Protocols and Why Does the Comparison Matter?
An RWA protocol takes an off-chain asset — a Treasury bill, a corporate loan, a fund share — and issues an on-chain token that represents a claim on it. The reason a side-by-side comparison is worth doing is that protocols filed under one label are running four different businesses with four different failure modes.
RWA Protocol Definition
The label covers issuers that hold the underlying asset and pass through its cash flow, credit markets that originate loans and take the credit risk, infrastructure that wraps or routes assets other people issued, and chains built to host regulated issuance. Only the first two hold anything on a depositor's behalf. The distinction matters because it determines what happens when something goes wrong: an issuer's failure is a custody and legal question, a credit market's failure is a recovery question, and an infrastructure failure is a downtime question. Comparing a Treasury issuer to a yield-trading layer on the single axis of total value locked treats four unlike things as one.
Comparison Criteria
Five axes separate these protocols usefully. What the protocol actually holds, measured on a consistent basis. Where the yield comes from, since a Treasury coupon and a private-credit spread are not the same promise. What the governance token entitles a holder to, which in most cases is a vote and nothing more. Who is eligible to participate, which frequently eliminates a reader before any other criterion applies. And what the protocol's record is under stress, which in 2026 is no longer a hypothetical for any of them.
Data current as of August 2026.
The taxonomy sets the frame; the sizes underneath it moved sharply in the past year.
How Big Is the On-Chain RWA Market in 2026?
The on-chain RWA market holds $38.32B in distributed asset value in August 2026, excluding stablecoins, across 2,522,832 asset holders (rwa.xyz, 2026-08-22). Both numbers are far above where this comparison found them a year ago, and the holder count in particular has changed character.
Sector Size Today
The $38.32B measures value actually distributed to on-chain holders, which is the honest denominator; the represented asset value on the same dashboard is $345.02B and counts assets registered rather than held. Holders grew to 2.52M from roughly 910,000, a jump driven less by capital than by a wave of small tokenized-equity positions — the registry now lists individual stock tokens with balances in the low thousands of dollars alongside billion-dollar funds. That composition change is worth holding onto when reading holder growth as an adoption signal. Tokenized Treasuries remain the largest category at $16.21B, followed by private credit at $7.25B, commodities at $4.92B, and tokenized stocks and ETFs at $2.33B.
Why Protocols Diverged
The past twelve months separated these six protocols more than the previous three years did. The two that grew — Ondo and Maple — both sell a product where the underlying asset is high-quality, the cash flow is contractual, and the buyer is institutional. The three that shrank did so for unrelated reasons: Pendle depended on an incentive programme that ended, Goldfinch depended on emerging-market borrowers who did not repay, and MANTRA depended on a token price that collapsed before its infrastructure thesis could be tested. Centrifuge held roughly flat while quietly changing what it does. Divergence of this size inside one sector label is the strongest available argument against treating "RWA protocol" as a category you can allocate to.
Reading the Field
Mapping protocols by what they hold, on a consistent measure, prevents most of the misleading conclusions. Pendle's TVL is not comparable to Ondo's because Pendle wraps yield tokens issued elsewhere and double-counts assets already counted by the issuer. Maple's figure depends heavily on which measure is quoted: the protocol reports $4.65B in assets under management, DefiLlama shows $2.761B in TVL, and rwa.xyz attributes roughly $954M in distributed value to it. All three are defensible and they are not interchangeable. Any comparison that does not say which measure it used is not a comparison.
Data current as of August 2026.
What Makes Ondo Finance the Tokenized Treasury Market Leader?
Ondo Finance holds about $3.49B in total value locked and roughly $3.77B in on-chain assets under management across more than 570 live assets, which makes it the largest pure-play tokenization platform by either measure (DefiLlama, 2026-08-22). What separates it from the rest of this list in 2026 is less the size than what it did with regulation.
Ondo Treasuries
Two products carry most of the balance. USDY, a yield-bearing note backed by short-term Treasuries and bank deposits, reached $2.15B in total asset value across twelve networks with 15,604 holders and a 7-day APY of 3.49%, having more than doubled since the start of 2026. OUSG, the fund for qualified purchasers, holds $366.93M, the significant majority of it in BlackRock's BUIDL with the remainder in a BlackRock money-market fund, bank deposits, and USDC for liquidity. Ondo's revenue reflects the scale: $13.26M in Q1 2026 and $15.3M in Q2. Worth noting for anyone reading revenue as token upside is that DefiLlama recorded gross profit of zero for those quarters, meaning the yield was passed through rather than retained.
USDY and OUSG Access
Access is the constraint, and it moved in 2026. USDY is open to non-US persons from a low minimum across twelve chains. OUSG is restricted to qualified purchasers with on-chain whitelisting. The change came through Ondo's broker-dealer subsidiary: after acquiring Oasis Pro in October 2025, Ondo secured FINRA authorizations on 23 July 2026 to offer tokenized equities and funds to US institutional and retail investors under SEC and FINRA oversight. That does not open OUSG or USDY to US persons, but it does mean the flat statement that Ondo has no US retail pathway — accurate for most of the protocol's life — is no longer true of the equities business.
How Do Maple Finance and Goldfinch Approach On-Chain Private Credit?
These two protocols ran the same experiment with opposite results, and 2026 delivered the verdict on both. Maple is the largest private-credit book in DeFi. Goldfinch voted to wind itself down.
Maple Credit Model
Maple runs an institutional credit market where pre-screened borrowers draw from pools against posted collateral, with overcollateralization doing the work that borrower selection alone did in the protocol's first design. It closed H1 2026 with $4.65B in assets under management, up 81% year over year, through a period when total DeFi value locked fell roughly 38%. The product line is four pools: syrupUSDC at $2.71B yielding 4.8%, syrupUSDT at $831M yielding 4.3%, syrupUSDG at $261M yielding 5.0%, and Maple Institutional at $849M yielding 5.2%. syrupUSDG is the notable addition, launched on Robinhood Chain as the collateral behind Robinhood Earn and reaching $200M in eight days. Trailing-twelve-month revenue is $22.09M against a net interest margin of 0.79%, and SYRUP holders approved a rules-based buyback in MIP-021 that directs 10% of monthly revenue to repurchases below $1.5M of monthly revenue, scaling to 30% above $2M. The first buybacks under that framework began in August 2026.
Goldfinch Emerging-Market Lending
Goldfinch lent to emerging-market borrowers without on-chain collateral, using Backers who supplied first-loss capital and Senior Pool depositors who funded the rest at lower yield. The model is now closed. Warbler Labs, the core development team, posted GIP-87 on 12 June 2026 proposing an orderly wind-down of Goldfinch Prime and a move to maintenance mode; the Snapshot vote closed on 23 June with about 1.05 million GFI cast, all in favour, against a 250,000 GFI quorum. Prime investors were redeemed in full and the contract was paused. Recovery of the legacy pools passes to a new US trust with the protocol's chief restructuring officer as trustee, and Warbler receives a fixed $150,000 for wind-down services over the next two or more years.
Default Track Record
The numbers behind that vote are the most instructive data point in this comparison. Goldfinch originated roughly $100M over its life. DefiLlama shows $56.15M in borrowed capital still outstanding against $1.63M in TVL, meaning close to every deposited dollar remains tied up in loans. One depositor account describes eight borrowers, two in default and six in restructuring, a realised loss rate near 70% against a protocol dashboard that had displayed roughly 20%, and withdrawals stalled for nearly three years. Maple's record is the counter-case: its 2022 losses came under an undercollateralized model, and the redesigned overcollateralized structure has run without credit losses since. The contrast is not really institutional versus emerging-market borrowers. It is whether the protocol held collateral it could actually seize.
Both records are public on-chain, which is the one genuine advantage this sector holds over private credit reporting.
How Does Centrifuge Connect Real-World Asset Originators to DeFi Liquidity?
Centrifuge holds $1.6B in TVL, down 6% quarter over quarter, with 10,988 CFG token holders, up 16% (Centrifuge Q2 2026 recap). The more useful observation is that the protocol answering this question today is not the one that answered it two years ago.
Centrifuge Origination
The original model connected asset originators — invoice financers, trade-finance desks — to DeFi capital through pooled receivables. The business that carries the balance now is tokenized institutional funds. The Janus Henderson Anemoy Treasury Fund, JTRSY, holds roughly $882M, carries an S&P fund credit-quality rating, reports a 30-day annualised yield near 3.24%, and requires a $500,000 minimum with a 0.25% management fee. Alongside it sit JAAA, a tokenized CLO strategy that took a $200M Ethena allocation on Solana, and Apollo's ACRDX. Origination from small private originators has not disappeared, but it is no longer where the money is.
Tranching and Pools
Centrifuge's structural innovation for 2026 is the deRWA wrapper rather than the senior/junior tranche. deJTRSY, deJAAA, and deCRDX are freely transferable on-chain wrappers of permissioned funds, which lets an asset requiring a $500,000 minimum and a whitelist circulate in DeFi as ordinary collateral. deSPXA gained traction on Uniswap and was accepted as collateral on Euler; the core assets went live on Monad. Centrifuge also partnered on a $1B redemption facility for JTRSY offering committed round-the-clock USDC redemptions rather than a queue. Senior and junior tranching still exists in the legacy pool structure and still allocates losses first to junior holders, but describing Centrifuge primarily as a tranching protocol now understates what it moved to.
What Do Pendle and MANTRA Offer That Other RWA Protocols Do Not?
Neither protocol holds real-world assets for depositors. Both are infrastructure plays, and both had a hard year — one from the unwinding of an incentive programme, the other from an exploit.
Pendle Yield Split
Pendle's mechanism is unchanged and still the cleverest in DeFi: any yield-bearing token splits into a Principal Token redeemable for the underlying at maturity and a Yield Token carrying all yield generated until then. Buy the PT at a discount and you have locked a fixed return; buy the YT and you have a leveraged position on the yield rate. What changed is the scale. TVL stands at about $1.165B, down roughly 91% from a $13.4B peak in September 2025, as the Ethena incentive cycle that had inflated the deposit base ended (DefiLlama, 2026-08-22). Pendle remains the largest yield protocol tracked by DefiLlama at 25.5% of that category, and it is now betting on a pivot toward RWA and institutional markets. The token mechanism also changed: sPENDLE replaced vote-escrowed PENDLE in January 2026, directing 80% of revenue to buybacks and cutting emissions by 77%, though a 2% perpetual annual inflation began in April 2026. A concentration risk sits underneath all of it, since a significant share of remaining deposits is still in Ethena-linked pools.
MANTRA Compliance Chain
MANTRA's thesis was a permissioned Layer 1 where compliance lives at the chain level rather than in each token contract. The thesis has not been disproven; it also has not been tested, because the protocol spent 2026 dealing with everything else. The OM token was redenominated to MANTRA on 2 March 2026 with a non-dilutive 1:4 split raising the hard cap to 10 billion, after the ERC-20 supply was deprecated in January. Market capitalisation is now about $21.8M, against roughly $226M for OM in June 2026 (CoinPaprika, 2026-08-22). Inveniam Capital Partners, which invested $20M in 2025, announced in June 2026 that it would acquire the project, with the deal expected to close in the third quarter, following January layoffs and restructuring. Then on 20 August 2026 an attacker exploited a vulnerability in the chain's Cosmos-EVM module and MANTRA halted its own chain as a precaution, freezing all endpoints and transactions, with exchanges suspending deposits and withdrawals and at least one placing the token on a caution list. Validators restarted the network on 22 August at 05:30 UTC on a patched release, v8.4.0, after roughly thirty hours of downtime and without rolling back state, meaning the attacker's transactions stand. Two MANTRA-managed wallets were drained. A full post-incident report had not been published at the time of writing, so the amount lost and the reimbursement position both remain open questions.
How Do RWA Protocol Tokens Perform Relative to TVL and Market Cap?
The disconnect between what these protocols hold and what their tokens are worth is the single most consistent finding in this comparison, and it held through a year in which the underlying assets grew.
TVL vs Market Cap
The ratio of assets managed to token market cap says something about value capture, though not what people usually assume. Ondo manages about $3.49B against a $1.80B market cap, a ratio near two. Maple's $2.76B of TVL sits against a SYRUP market cap in the low hundreds of millions. Centrifuge holds $1.6B against a $79.5M CFG market cap, a ratio above twenty and the widest in this group. A high ratio can mean the token is cheap relative to the business or that the token captures almost nothing the business earns, and across this sector the second reading has been right far more often. Pendle is the instructive case for the opposite reason: its ratio compressed not because the token rose but because the TVL fell 91%, which is a reminder that these ratios move on the denominator.
Token Value Capture
The mechanisms genuinely differ. SYRUP now has the clearest link, with MIP-021 directing 10% to 30% of monthly revenue to buybacks depending on revenue level, and the first repurchases executed in August 2026. sPENDLE routes 80% of revenue to buybacks, partially offset by 2% annual inflation. CFG accrues through network fees and governance. ONDO carries governance rights with limited direct fee exposure at current settings, which is the gap that explains a $1.80B market cap against a business collecting $15.3M in a quarter. GFI votes on a protocol in wind-down. Where a token has no revenue claim, growth in assets under management reaches holders only through sentiment.
Supply Dilution
Circulating supply relative to maximum is the other half of the price story, and ONDO's moved materially. 4.869B ONDO now circulate against a 10B maximum, or 48.7%, up from 31.6% in June 2026 — roughly 1.7B tokens entered circulation in a year (CoinPaprika, 2026-08-22). PENDLE circulates 163.8M of 258.4M, or 63.4%, unchanged, though the new 2% perpetual inflation changes the trajectory. CFG has no stated maximum. GFI circulates 29.1M of 114.3M. Dilution alone does not determine price, but a token releasing supply on a schedule into a business that pays holders nothing has two forces pushing the same direction.
Data current as of August 2026.
One caveat on the MANTRA row: the March 2026 redenomination restarted the price series, so the drawdown shown is measured from the post-split high, not from the pre-crash OM peak, which was very much steeper.
How Do RWA Protocols Compare on Yield Potential and Credit Risk?
The headline change here is that the yield spread between the safest and riskiest products in this comparison has nearly closed. The old framing of 3–5% Treasuries against 8–15% private credit described a market that no longer exists.
Yield vs Risk
Ondo's USDY publishes a 7-day APY of 3.49%; Centrifuge's JTRSY reports a 30-day annualised yield near 3.24%. Maple's four pools run from 4.3% on syrupUSDT to 5.2% on Maple Institutional. That is the entire practical range for anyone comparing these protocols today: roughly 3.2% to 5.2%, or about two percentage points from a highly rated Treasury fund to an institutional credit book. The risk difference across that span is not two percentage points' worth. A tokenized Treasury fund carries duration and regulatory-access risk; an institutional credit pool carries borrower default risk that Goldfinch just demonstrated is not theoretical. Pendle's Yield Tokens remain the exception with genuinely variable returns, but a YT is a leveraged bet on a rate, not a yield product, and it can expire worthless without any default occurring.
Credit vs Treasury Risk
The useful lesson from 2026 is about what backs the claim rather than what the yield number says. Maple's overcollateralization means a defaulting borrower's collateral can be seized and sold, which is why its losses in a defaulting cycle were recoverable and Goldfinch's were not. Goldfinch's uncollateralized emerging-market model produced a realised loss rate reported near 70% against a dashboard showing roughly 20% — a gap between displayed and actual risk that no yield premium compensated for. Centrifuge's senior tranche holders still sit behind junior capital, and its newer fund products carry the credit quality of the underlying manager rather than of a pool of small originators. When the spread between products is two points, the structure of the claim matters far more than the rate.
Data current as of August 2026.
The table shows no product offering a high yield at low risk, which is the same finding as last year — but it now also shows almost no product offering a high yield at all.
Which RWA Protocol Best Matches Different Investor Types and Goals?
Selection reduces to three questions: what you are eligible for, what you want the money to do, and whether the protocol will still be operating when you want it back. The third question was not on this list a year ago.
Match to Investor Type
Conservative allocators wanting Treasury-grade yield map to Ondo's USDY at a low minimum outside the US, or to OUSG if they clear the qualified-purchaser bar. Institutional allocators wanting a credit spread map to Maple, where the overcollateralized structure, the published pool-level reporting and the buyback mechanism are the closest thing in DeFi to an institutional credit product. Allocators who want rated fund exposure usable as DeFi collateral map to Centrifuge's deRWA wrappers, which is a genuinely new capability. Traders with a view on rates map to Pendle. Two categories that appeared in earlier versions of this comparison no longer have a clean match: emerging-market credit exposure, which Goldfinch's wind-down closed, and regulated-chain infrastructure exposure, where MANTRA is running again after its August exploit but has not yet published a post-incident report or closed its acquisition.
Diversification Logic
Spreading across protocols still helps, but the case is weaker than it was when the yield spread was ten points wide. Combining a Treasury product with a credit product now buys roughly 1.5 percentage points of additional yield in exchange for a genuinely different failure mode, and each additional protocol adds its own eligibility check, minimum, and redemption schedule. The stronger argument for diversification in 2026 is not yield optimisation but operator risk: Goldfinch's depositors were not harmed by holding the wrong asset class, they were harmed by holding all of it in one protocol whose borrowers defaulted and which is now in wind-down.
Data current as of August 2026.
Eligibility eliminates more readers than any other criterion in this table, so check it before comparing anything else.
What Regulatory and Market Catalysts Will Shape RWA Protocols in Late 2026?
One catalyst that was speculative in earlier versions of this comparison has now happened, and it is larger than the version that was predicted.
Regulatory Catalysts
On 15 July 2026 the Depository Trust & Clearing Corporation processed real production trades using tokenized representations of securities held at The Depository Trust Company, with more than 30 firms participating, including BlackRock, JPMorgan, Goldman Sachs, Nasdaq, the New York Stock Exchange, Vanguard, and Ondo Finance. The DTCC Tokenization Service is scheduled to launch in October 2026, roughly seven months after the SEC issued DTC a no-action letter authorizing it. Ondo used the service to issue the first tokenized stocks backed by DTC tokenized entitlements, carrying the same CUSIP and symbol as the underlying. Eight days later, on 23 July, its broker-dealer subsidiary received FINRA authorizations to offer tokenized equities and funds to US institutional and retail investors. Post-trade infrastructure integration was the thing this sector had been waiting for, and it arrived through the incumbents rather than around them.
What to Watch
Four indicators will say more than TVL headlines. Whether the DTCC Tokenization Service ships on schedule in October and how many of the 30-plus participants transact on it. Whether Maple converts its $4.65B and its buyback mechanism into a token price that reflects the business, which is the cleanest live test of whether RWA governance tokens can capture value at all. Whether Pendle's pivot to RWA and institutional pools rebuilds a deposit base that fell 91%. And whether MANTRA, whose chain restarted on 22 August after roughly thirty hours down, publishes a full post-incident report and completes the Inveniam acquisition — two conditions still unresolved.
The sector's $38.32B and 2.52 million holders in August 2026 are a real foundation. The last twelve months also showed that a protocol inside it can lose 91% of deposits, halt its own chain, or vote to close, without the sector total moving much at all.
Summary
RWA protocols tokenize off-chain assets and issue on-chain claims against them, but the six covered here run four different businesses. Ondo issues tokenized Treasuries and equities and holds the asset. Maple and Goldfinch originate credit and take the credit risk, though Goldfinch is now in wind-down and closed to new capital. Centrifuge tokenizes institutional funds and wraps them for DeFi use. Pendle trades yield generated by other protocols' assets, and MANTRA builds a chain for regulated issuance. Access requirements differ at every step and frequently decide the question before yield or risk criteria apply.
The year separated them sharply. Ondo holds about $3.49B in TVL and $3.77B in on-chain assets under management, with USDY at $2.15B yielding 3.49% and OUSG at $366.93M; its broker-dealer subsidiary received FINRA authorizations in July 2026 to serve US investors with tokenized equities. Maple reports $4.65B in assets under management, up 81% year over year, across four pools yielding 4.3% to 5.2%, with revenue-linked buybacks live since August 2026. Centrifuge holds $1.6B, down 6% quarter over quarter, having pivoted from originator pools to tokenized institutional funds such as the $882M JTRSY, and now wraps them as freely transferable deRWA collateral. Pendle fell to about $1.17B from a $13.4B peak, a 91% decline. MANTRA's token trades at a $21.8M market cap after redenomination, an acquisition announcement, and an August exploit that halted its chain for roughly thirty hours before a patched restart on 22 August. Goldfinch voted on 23 June 2026 to wind down, with roughly $56M outstanding against $1.63M of TVL and a stated recovery horizon of two or more years. Across all six, governance tokens sit 78% to 99% below their highs while the sector grew to $38.32B — and the yield range that once justified taking credit risk has compressed to roughly 3.2% to 5.2%.
Conclusion
The comparison that mattered a year ago was which protocol offered the best yield for the risk. The comparison that matters now is which protocols are still operating on the terms they advertised. Two of the six grew through a shrinking DeFi market by selling contractual cash flow to institutional buyers. One pivoted its business quietly and successfully. One lost nine tenths of its deposits when someone else's incentive programme ended. One halted its own blockchain. One voted to close after borrowers stopped paying, leaving depositors with a two-year recovery horizon and a realised loss rate reported near 70%. TVL rankings would have told you almost none of that, which is the argument for reading what a protocol holds, who it lends to, and what happens to your claim if it stops.
Why You Might Be Interested?
If you hold a position in any of these protocols, the sections on Goldfinch's wind-down and MANTRA's exploit describe two failure modes that arrived in the past ten weeks. If you are comparing yields, the yield section shows the whole range now fits between roughly 3.2% and 5.2%, which changes the case for taking credit risk. And if you hold a governance token expecting it to track the sector, the token section quantifies how completely that relationship has broken.
Quick Stats
- $38.32B — tokenized real-world assets on-chain in August 2026, excluding stablecoins, across 2,522,832 holders
- $4.65B — Maple Finance assets under management, up 81% year over year while total DeFi value locked fell roughly 38%
- -91% — Pendle's TVL decline from a $13.4B peak in September 2025 to about $1.17B
- $56.15M — borrower capital still outstanding at Goldfinch, now in wind-down, against $1.63M of TVL, with recovery estimated at two or more years
- $21.8M — MANTRA's market cap in August 2026, against roughly $226M for the predecessor OM token in June
- 3.2% to 5.2% — the entire yield range across the operating protocols in this comparison
Data current as of August 2026.
FAQ
?What is the difference between RWA protocols and stablecoins?
A stablecoin holds a peg to a reference currency and pays its holder nothing; the issuer keeps the yield on the reserves. An RWA protocol issues a token whose value derives from a yield-generating off-chain asset and passes some of that yield through. Ondo's USDY, for example, published a 7-day APY of 3.49% in August 2026 sourced from short-term Treasuries and bank deposits, while a USDC holder earns nothing on the stablecoin itself unless they deposit it somewhere else.
?Is Pendle still the largest RWA-adjacent protocol by TVL?
No. Pendle's TVL fell about 91% from a $13.4B peak in September 2025 to roughly $1.165B as the Ethena incentive cycle unwound, putting it behind Ondo, Maple, and Centrifuge. It remains the largest yield-trading protocol tracked by DefiLlama at about 25.5% of that category. Its TVL was also never directly comparable to theirs, since Pendle wraps yield tokens issued by other protocols and double-counts assets already held elsewhere.
?What happened to Goldfinch?
Goldfinch voted to wind down. Warbler Labs posted GIP-87 on 12 June 2026, and the Snapshot vote closed on 23 June with about 1.05 million GFI in favour and none against. Goldfinch Prime was shut and its investors redeemed in full; the wider protocol moved to maintenance mode focused only on collecting legacy borrower payments. Roughly $100M was originated over the protocol's life, about $56.15M remains outstanding against $1.63M of TVL, and the stated recovery horizon is two or more years. Recovery passes to a US trust overseen by the protocol's chief restructuring officer.
?Is Maple Finance safe after its 2022 defaults?
Maple's 2022 losses occurred under an undercollateralized model that the protocol replaced. Under the current overcollateralized design, borrowers post collateral exceeding the loan, and the protocol has run without credit losses since. It closed H1 2026 with $4.65B in assets under management, up 81% year over year, through a period when total DeFi value locked fell roughly 38%. The structural point Goldfinch demonstrated is that collateral you can seize is what separates a recoverable default from an unrecoverable one — but no lending protocol is risk-free, and collateral can fall in value faster than it can be sold.
?Can retail investors access these protocols?
Access varies sharply. Ondo's USDY is open to non-US persons at a low minimum across twelve chains; OUSG requires qualified-purchaser status. Maple's pools require KYC and are aimed at institutional tickets. Centrifuge's flagship fund requires a $500,000 minimum, though its deRWA wrappers trade freely once issued. Pendle has no accreditation requirement, so any wallet can trade PT and YT. Goldfinch is closed to new capital entirely, having wound down in June 2026.
?What does the ONDO token actually do?
ONDO grants governance rights over protocol parameters and treasury allocation. It does not entitle holders to a share of the management fees Ondo earns on OUSG, USDY, or Ondo Stocks at current settings. That is the gap behind a $1.80B market cap against a business that recorded $15.3M of gross protocol revenue in Q2 2026, and it sits alongside a supply structure that expanded from 31.6% to 48.7% of the 10B maximum over the past year.
?How does Centrifuge handle defaults in its pools?
Legacy Centrifuge pools use senior and junior tranches, and losses absorb into the junior tranche before reaching senior holders. Some pools entered workout in past cycles; in most documented cases senior investors recovered principal while junior investors took the loss, which is the tranche structure working as designed. The larger part of Centrifuge's book today is tokenized institutional funds such as JTRSY and JAAA, where the relevant credit question is the quality of the underlying portfolio and its manager rather than the tranche waterfall.
?What is the MANTRA token used for after its April 2025 collapse?
The token was redenominated from OM to MANTRA on 2 March 2026 in a non-dilutive 1:4 split that raised the hard cap to 10 billion. Its market cap is about $21.8M in August 2026. The compliance-chain thesis — hosting regulated issuance with identity verification enforced at the chain level — is still the stated purpose, but it took a hit in August 2026, when an attacker exploited a vulnerability in the chain's Cosmos-EVM module and MANTRA halted the network for roughly thirty hours before restarting on a patched release on 22 August without rolling back state. A full post-incident report had not been published at the time of writing. An acquisition by Inveniam Capital Partners announced in June 2026 was expected to close in the third quarter.
?Are RWA protocols regulated?
It depends on the product and the jurisdiction, and the answer changed in 2026. Ondo structures OUSG under a US private-placement exemption and USDY as an offshore note, and its Oasis Pro subsidiary now holds SEC and FINRA registrations covering tokenized equities and funds for US investors. Maple operates without a single primary regulator. Centrifuge's fund products sit inside conventional fund structures with rated portfolios and named administrators. MANTRA markets itself as a regulated Layer 1 under UAE and other frameworks. None of them operates under a single global umbrella, and a registration held by one subsidiary does not extend to a protocol's other products.
References / Sources
Market and protocol data
Current sizes, yields, and holder counts for the protocols compared.
- rwa.xyz: Global Market Overview, distributed asset value and holder count (rwa.xyz, Aug 2026)
- DefiLlama: protocol pages and income statements for Ondo Finance, Maple Finance, and Pendle (defillama.com, Aug 2026)
- Maple Finance: Transparency dashboard and Maple Memo, July 2026 (maple.finance, Aug 2026)
- Ondo Finance: USDY and OUSG portfolio pages (ondo.finance, Aug 2026)
- Centrifuge: Q2 2026 recap, TVL and holder metrics (centrifuge.io, 2026)
- CoinPaprika: market data and ATH distances for ONDO, PENDLE, CFG, MANTRA, MPL, GFI (coinpaprika.com, Aug 2026)
Governance and regulatory record
Primary documents behind the wind-down, exploit, and infrastructure developments.
- Goldfinch Governance: GIP-87, Maintenance Mode of Goldfinch Operations and Wind-Down of Goldfinch Prime (gov.goldfinch.finance, Jun 2026)
- The Block: Goldfinch set to shutter Prime after community vote backs wind-down proposal (theblock.co, Jun 2026)
- DTCC: DTCC Turns Tokenization into Reality, US trades processed using DTC-tokenized assets (dtcc.com, Jul 2026)
- Ondo Finance: Oasis Pro Markets secures FINRA authorizations (prnewswire.com, Jul 2026)
- CoinDesk: MANTRA token plunges to record low as blockchain halts after exploit (coindesk.com, Aug 2026)
Original CoinPaprika guide.