GSR's Hare Is a $100M Bet on Onchain Credit Curation, Not Yet Proof of Demand

Executive Summary

On October 7, 2026, GSR announced that it is committing $100 million, mostly through a credit facility, to Hare, a new onchain vault business built with the liquidity distribution platform Turtle (CoinDesk). Hare creates and manages vaults. Its first two products run on Aave. Our reading is narrow: Hare shows a large market maker will back credit curation with its own balance sheet. It does not yet show that outside institutions, let alone retail users, are ready to buy tokenized credit.

Three facts anchor that reading. The first two vaults were described as "coming soon", with opening dates and subscription terms to follow, and the amount of GSR capital already deployed was not disclosed (Cryptonews.net). The commitment is GSR's own capital, so it measures GSR's conviction, not outside demand. And GSR group CEO Xin Song said in a March 2026 podcast that "the demand side hasn't really come yet" (TokenPost). Separately, the IMF puts tokenized credit at $30.4 billion as of July 31, 2026 (Crypto.news).

The test for Hare is therefore operational. Can it attract depositors beyond GSR, disclose how its yields are produced, and manage redemption and credit risk when conditions tighten? This report sets out what is known, what is not, and what would change our view. It relies on press coverage and aggregator reproductions of Hare, GSR and IMF material, not on the primary documents, so Hare and IMF statements are attributed to the outlets that reported them.

Market Context

Hare launches into a large but familiar base of onchain dollar liquidity. Total stablecoin supply was $312.90 billion on October 13, 2026 (DefiLlama), with USDT at 58.8% and USDC at 23.3%. Hare USD Earn is designed to accept major dollar stablecoins in a single vault (PANews). The addressable pool is therefore large. That says nothing about the share Hare could capture.

Ethereum, the main settlement venue for Aave-type lending, held $51.58 billion in chain TVL on October 13 (DefiLlama). Aave protocol TVL was $18.65 billion. Morpho stood at $11.34 billion and Maple at $2.79 billion (all DefiLlama, October 13). The AAVE token traded near $158 with a market cap of $2.44 billion at 10:20 UTC (DefiLlama, CoinGecko id aave).

The wider tokenization market is now being sized by the IMF. Its October 2026 Global Financial Stability Report pairs size estimates with systemic-risk warnings, covered below.

What GSR Announced, and What It Did Not

GSR announced that it will commit $100 million to Hare, mostly through a credit facility (CoinDesk). A second report describes the commitment as multi-year and mostly in the form of credit lines (PANews). GSR's capital is deployed before external investors enter, and Paxos Labs is collaborating on the gold vault.

Four things were not disclosed. The first is the amount of GSR capital already deployed (Cryptonews.net). The second is vault opening dates and subscription terms, which were promised later. The third is yields and caps. The fourth is how the fixed-rate yield on Hare USD Earn will be generated (Cryptonews.net). Hare's launch messaging says fixed-rate yield is available, but the source of that rate is unstated.

The practical consequence is that Hare is currently a business announcement, not a product with TVL. There is no deposit base to measure and no yield history to evaluate. A multi-year credit line is not capital already at work, and the deployed amount is undisclosed.

The commitment is GSR's money

It is easy to misread the commitment as institutional demand. It is not. The capital belongs to GSR, mostly as credit lines. A market maker lending to its own vault business shows how GSR sees the opportunity, but it adds no information about whether outside allocators want the product.

Hare CEO Connor Milner told CoinDesk that the commitment is deployment capital and that "allocators see GSR's own capital in the same vaults as theirs" (Cryptonews.net, citing CoinDesk). That is a real alignment feature: GSR is not only collecting fees. It also means early vault balances will mix GSR's anchor liquidity with outside deposits. When Hare publishes vault data, the useful number is outside deposits net of GSR's share.

Aave is the venue, not the backer

Aave is the lending venue powering the vaults, and GSR is the funder (CoinDesk, PANews). The distinction matters for risk. Aave supplies the underlying market and its smart contracts. GSR supplies capital, underwriting and curation.

How the Hare Model Works

Per Hare's own launch messaging, reproduced by KuCoin News, Hare is a credit desk by GSR for onchain finance. It provides underwriting, structuring and vault curation for both issuers and allocators. The post says the launch is supported by Chainlink and Hypernative. This is an institution-oriented positioning: Hare USD Earn is described as built for larger allocations and to limit yield compression as deposits grow.

Two initial vaults, neither an explicit private-credit product

The first products are narrower than the "onchain credit desk" label suggests. Hare USD Earn accepts major dollar stablecoins in a single vault. Hare Gold Earn is for holders of Paxos' tokenized gold products PAXG and PAXGy. GSR provides anchor liquidity (PANews).

Neither is described in the sources as a private-credit product. That matters when readers compare Hare with the IMF's tokenized credit figure. The sources reviewed do not say whether or when Hare will add offchain credit exposure, although Milner's risk framework refers to offchain credit risk. For now the sourced description is of Aave-powered stablecoin and gold products, not of loans to offchain borrowers.

What "curation" has to deliver

A curation layer earns its place by doing three things that depositors cannot easily do themselves. It must select and size exposures. It must monitor them. And it must manage liquidity so that depositors can leave when they want to. Milner said in a GSR-published interview that Hare would assess underlying offchain credit risk plus smart-contract and market risks, and stress-test how quickly investors could get money back in a rush to redeem (Cryptonews.net). He also said that calling these products riskless "would be a lie".

That candour is useful. It also sets the standard Hare will be judged against. A credit desk that underwrites for both issuers and allocators sits on two sides of the market, and GSR also provides the capital. Whether those roles are separated by process or only by branding is not described in the sources reviewed. We treat it as an open question, not a finding.

Yield design is the largest gap

The stated design goal for Hare USD Earn is to limit yield compression as deposits grow. The sourced descriptions give no mechanism for how that is achieved. Hare has not detailed how fixed-rate yield will be generated (Cryptonews.net). A fixed rate could come from GSR's balance sheet, from the underlying venue, or from another structure. Each has different risk implications for depositors, and none is confirmed.

We found no quantified evidence of collateral-efficiency gains versus conventional DeFi lending, and no sourced figure on yield. Claims of efficiency should be treated as unproven until Hare publishes vault terms and performance.

Sizing the Category Depends on Definitions

The IMF's October 2026 Global Financial Stability Report estimates tokenized real-world assets at about $65.1 billion as of July 31, 2026, excluding repurchase agreements and stablecoins (Bitcoin.com News, reporting the IMF). Within that total, tokenized credit was $30.4 billion, money market funds $17.5 billion and tokenized equities about $2.3 billion (Crypto.news, reporting the IMF). By our computation, credit is about 47% of the IMF total. The remaining roughly $14.9 billion falls in categories not itemized in the sources we reviewed.

Three caveats apply before anyone uses these figures as a demand signal.

First, the IMF covers "publicly reported" tokenized RWAs, so it does not capture private or unreported issuance (CoinPaprika). The IMF also described the total as a small fraction of roughly $300 trillion in global capital-market assets.

Second, other trackers give different answers. TokenPost reported tokenized real-world assets above $34 billion as of August 31, from an unnamed dataset. That is a different and narrower measure than the IMF's, and the two should not be compared or combined. Crypto Briefing cited third-party estimates of total tokenized RWAs from $39 billion to $60 billion, and of the 2026 tokenized private credit market from $4 billion to $18.9 billion.

Third, the IMF's credit category is probably broader than private credit, and its definition was not confirmed in the sources reviewed. We therefore do not equate it with "tokenized private credit". The gap between the IMF credit figure and the private-credit estimates is consistent with that.

Scale: the commitment against the category

The commitment equals about 0.3% of the IMF credit figure. That is a scale comparison only. Hare's first vaults are stablecoin and gold products, the commitment is mostly credit lines with an undisclosed deployed amount, and the IMF category is defined differently.

We do not publish a 2027 market-size estimate. None was found in a reputable source, and projecting one would add a number the evidence cannot support.

The Demand Problem

The most important data point for Hare is not the commitment. It is the demand gap that GSR's own CEO described. Xin Song said on a podcast published March 26, 2026: "There's a lot of people that are supplying assets, but the demand side hasn't really come yet" (TokenPost). Media in October 2026 tied the remark to Hare. It is about six months older than the launch and was not made in response to it.

Read in that light, Hare can be seen as an attempt to address the problem Song described. If supply is ahead of buyers, a curation layer that pairs issuers with allocators and puts a market maker's capital in the same vaults is a plausible way to build the demand side. It is also a bet that the gap is about trust, structuring and distribution, not about price or fundamentals. The evidence does not yet say which.

Supply growth is not demand

Crypto Briefing noted that many tokenized RWAs record zero trades in a given week. Issuance can grow while secondary liquidity stays thin. The IMF's trading data point the same way: tokenized repos dominate tokenized trading activity at about $300 billion to $350 billion of average daily volume, versus roughly $13 trillion traded daily in the broader US repo market (Cointelegraph, citing the IMF).

That puts tokenized repo volume at roughly 2% to 3% of the broader daily market, by our arithmetic. More important is the concentration. Trading is dominated by repo, a short-dated, collateral-driven product, not by longer-dated credit. A vault holding credit-like exposure cannot assume a deep secondary market will be there to absorb exits.

The curated-vault base

Hare is entering an existing category. As of July 2026 there were $8.6 billion in assets across 788 curated vaults, per Vaults.fyi data cited by CoinDesk. That is roughly $11 million per vault on average, by our computation. The average alone does not show how assets are distributed across vaults.

The category shows that depositors will use curated vaults. It does not show how many of those assets sit in vaults with credit risk like what Hare describes. It also leaves open whether larger allocations of the kind Hare targets will route through new entrants or through established curators.

A comparison point: Dow Protocol

Another institutional vault shows what a credit-oriented product disclosure looks like, and how early the field is. Dow Protocol announced an institutional vault as "coming soon", not launched. It targets short-duration, self-liquidating e-commerce credit and is backed by an overcollateralized receivables pool with at least 120% coverage plus structured repayment controls (Dow Protocol post, reproduced by KuCoin News). The coverage figure is the issuer's own claim.

Hare's first two vaults are different. They are stablecoin and gold products. If Hare moves into offchain credit, disclosures of the Dow type, covering collateral coverage, repayment controls and duration, will be the standard against which allocators judge it.

Where Hare Sits Among Onchain Credit Venues

Aave, Morpho and Maple are useful context for Hare's competitive position, not a like-for-like ranking. They differ in design and in the credit exposure they carry. By TVL on October 13 (DefiLlama), Aave is about 1.6 times Morpho and about 6.7 times Maple. Morpho is about 4.1 times Maple.

Hare's choice of Aave as its first venue has two implications. It gives the vaults access to the largest of the three lending pools by TVL, which helps with liquidity. It also means Hare's products inherit that venue's smart-contract and market risks, which Milner's framework lists explicitly.

Against these venues, the commitment looks modest. It is about 0.5% of Aave's TVL and about 1.2% of the July curated-vault base, though the dates differ and the commitment is not deployed capital. Hare's differentiation, if any, would have to come from underwriting quality, distribution through Turtle and trust in GSR's risk management rather than balance-sheet scale.

Nothing in the sources links Hare to a price move in the AAVE token. We do not treat the launch as a token catalyst.

Redemption Stress and the Plumbing Gaps

The IMF's warnings describe risks that also apply to curated credit vaults. As summarized by ChainCatcher, the IMF warned that automatic liquidations, instant settlement, continuous trading and programmable collateral could accelerate forced liquidations and amplify financial shocks. It also flagged cross-platform collateral reuse as a risk.

For a vault, the mechanism is straightforward. Programmable systems act faster than human credit committees. If asset prices fall or redemptions rise together, automated rules can force sales into thin markets. Where the same collateral backs positions on several platforms, stress on one can spread to others.

Three named obstacles

The IMF named obstacles to tokenized-market growth that include legal uncertainty, insufficient interoperability and the lack of a widely accepted settlement asset (KuCoin News; CoinPaprika). CoinPaprika reports four interdependent constraints in total. The sources we reviewed name only these three.

Each bears on Hare in a specific way:

• Legal uncertainty affects what an allocator can enforce against an offchain borrower if a credit goes bad.
• Interoperability limits how easily a vault position or its collateral can move between platforms.
• Settlement asset concerns what an allocator accepts as final payment, which stablecoin-denominated vaults partly answer but do not resolve.

Milner's own stress test

Milner's statement that Hare will stress-test how quickly investors could get money back in a rush to redeem is the right instinct. The sourced materials do not describe withdrawal terms, notice periods or liquidity buffers for the first two vaults. Until they do, the claim cannot be assessed.

A reasonable standard is that vault terms match the liquidity of the underlying assets. A vault offering instant redemption against assets that rarely trade is a mismatch. Whether Hare avoids it will be visible in its subscription and redemption terms once published.

What to Watch Next

Five observable items would move our assessment.

1. Vault opening and terms. Dates, caps, subscription terms and withdrawal mechanics for Hare USD Earn and Hare Gold Earn.
1. Deployed capital. How much of GSR's commitment is drawn, and how much of each vault is GSR versus outside depositors.
1. Fixed-rate source. Whether the fixed-rate yield is produced by the underlying venue, by GSR's balance sheet or by another structure.
1. Eligibility. Who may deposit, and whether KYC or eligibility gates apply. All sourced positioning is institutional, around larger allocations. No sourced evidence shows a retail pathway, so we treat retail integration as an open question.
1. Credit expansion. Whether Hare adds vaults with offchain credit exposure, and what coverage, duration and repayment disclosures it provides.

Data: Sizing and Venue Comparison

The first table collects the competing size estimates for tokenized assets. The IMF figures are as of July 31, 2026, as reported by the outlets named. The other rows carry their own dates and sources and should not be added together.

The second table gives the onchain venue context. All figures are from DefiLlama on October 13, 2026.

Two cautions apply to the tables. The IMF rows measure stock at one date, not live activity. The venue rows measure deposits, not credit quality or loan performance, so a larger TVL does not imply lower risk.

Risks

Ratings are our judgment of likelihood and impact over the next three to six months.

• Demand shortfall: High. The vaults open but attract few outside depositors, leaving GSR's anchor liquidity as most of the balance. Trigger: published vault data showing GSR as the dominant depositor several weeks after opening, consistent with the demand gap Song described.
• Redemption stress: Medium. Withdrawal demand rises faster than underlying liquidity allows. Trigger: a market drawdown or a credit event while vault terms promise faster exits than the assets can deliver. Milner has said Hare will stress-test this, but no terms are public.
• Disclosure gaps: Medium. Terms, deployed capital and the source of fixed-rate yield stay undisclosed after launch. Trigger: vaults open without published caps, yield mechanics or GSR's share of deposits. Allocators then cannot price the risk.
• Concentration and role conflict: Medium. GSR funds, underwrites and curates, and Hare serves both issuers and allocators. Trigger: a credit decision that favors an issuer relationship over depositors, or capital allocation that shows no separation of roles.
• Smart-contract and venue risk: Medium. Hare's vaults inherit exposure to Aave's contracts and markets, a risk Milner's own framework lists. Trigger: an exploit or severe market disruption at the underlying venue.
• Collateral reuse and liquidation cascades: Medium for Hare, High for the system. The IMF flags programmable collateral and cross-platform reuse as amplifiers. Trigger: stress that forces automated liquidations across platforms holding the same collateral.
• Legal and settlement gaps: Medium. Enforcement against offchain borrowers is uncertain and no widely accepted settlement asset exists. Trigger: Hare adds offchain credit exposure before legal structures are tested.
• Data misinterpretation: Low to Medium. Market-size figures vary by definition, and the IMF and TokenPost numbers measure different things. Trigger: commentary that adds or compares the figures as if they shared one definition.

Outlook

Our three to six month view runs from now to roughly the first quarter of 2027. The probabilities below are our estimates, based on the thin disclosure so far and on the demand gap described by GSR's CEO. They are not sourced facts.

Base case (55%). The first vaults open and publish terms, but visible outside deposits stay modest and GSR's anchor capital remains a large share of balances. Hare's fixed-rate yield source and eligibility rules are partly clarified. The market reads Hare as a credible but early curation business. This is the most likely outcome because a launch announced as "coming soon" has to clear several disclosure steps, and the demand gap is a feature of the market, not of one product.

Upside case (25%). The vaults open on schedule with clear terms, outside depositors participate meaningfully alongside GSR, and Hare discloses the source of its fixed-rate yield. Hare then adds a vault with offchain credit exposure and Dow-style collateral disclosure. This would be the first evidence that a market maker's curation layer can create demand, not just reflect it.

Downside case (20%). Vault openings slip or terms stay opaque, or outside deposits are negligible. A market shock could also expose a mismatch between redemption terms and asset liquidity. The result would confirm the supply-ahead-of-demand reading and weaken the case for curated credit as an institutional product.

Takeaways

For traders: Hare is not a tradable instrument, and nothing in the sources links it to the AAVE token. Treat the launch as an information event about GSR's strategy, not a demand signal. Watch for published vault data, such as the curated-vault data Vaults.fyi tracks.

For builders: The standard for credit vaults is set by disclosure. Terms that state collateral coverage, duration, repayment controls and redemption mechanics will matter more than headline commitments. Builders targeting institutional allocators should match vault liquidity to underlying asset liquidity and say so plainly.

For investors: Separate GSR's capital from outside capital whenever vault TVL is reported. Do not use the IMF's tokenized-credit figure as a proxy for private-credit demand, because its definition is unconfirmed. Treat any retail-access, collateral-efficiency or 2027 market-size claim about Hare as unsupported until a primary source states it.

Sources

• CoinDesk: GSR's Hare launch, $100 million commitment, curated-vault base
• PANews: Hare vaults, GSR anchor liquidity, multi-year credit lines, Paxos Labs
• Cryptonews.net: vaults "coming soon", undisclosed deployed capital, Milner statements, fixed-rate gap
• KuCoin News: Hare launch messaging, Chainlink and Hypernative support
• Bitcoin.com News: IMF total tokenized RWA estimate
• Crypto.news: IMF breakdown of credit, money market funds and equities
• CoinPaprika: IMF "publicly reported" scope and constraints
• ChainCatcher: IMF warnings on liquidations and collateral reuse
• KuCoin News (ME News): IMF obstacles to tokenized-market growth
• TokenPost: Xin Song podcast remarks and the $34 billion tokenized RWA figure
• Crypto Briefing: range of tokenized RWA and private credit estimates, thin trading
• KuCoin News: Dow Protocol institutional vault announcement
• Cointelegraph: IMF on tokenized repo volume versus the US repo market
• DefiLlama: Aave protocol TVL
• DefiLlama: Morpho TVL
• DefiLlama: Maple TVL
• DefiLlama: Ethereum chain TVL
• DefiLlama Stablecoins: total stablecoin supply and shares
• CoinGecko via DefiLlama: AAVE price and market cap