HTX Ventures Latest Report | The Convergence of CeFi, DeFi, and TradFi in 2026: How Institutional-Grade Hybrid Architecture Is Taking Shape

Introduction

If you open the interface of any leading custody platform in 2026, you are likely to see a new button: “Allocate idle stablecoins to an institutional yield strategy.” Clicking it routes your money through the custodian, a risk curator, and a Vault infrastructure layer before it is deployed into Aave or Morpho lending markets, all without ever touching a wallet, signing a transaction, or researching a single protocol. Meanwhile, at the other end, a hedge fund can keep tokenized money market fund shares issued by Franklin Templeton with its custodian while using them as margin on Binance, allowing those shares to continue earning Treasury income.

---

Though these appear to be two unrelated product innovations, they point to a single change: as institutions adopt digital assets, the five core functions—custody, collateral, execution, yield, and risk management—are decoupling from all-in-one platforms into a modular network of specialized players. This article calls this shift the move from Platform-Based Finance to Modular Institutional Finance, and attempts to answer three questions: why this unbundling happened in 2026; who owns each function and how profits are split after the break-up; and whether this is a few isolated product integrations or a longer-term shift in market structure.

This report is prepared by HTX Ventures, the global investment arm of HTX. Through its investment and research activities, HTX Ventures continuously tracks the institutionalization of custody, collateral, trade execution, and on-chain yield, with a focus on how institutional capital flows across these functions and how returns and risk responsibilities are allocated. The analysis in this report is based on publicly disclosed partnership arrangements, on-chain data, and regulatory filings.

I. The Core Shift in 2026: From Product Integration to Financial Architecture Restructuring

Three Parallel Tracks

From 2023 to 2025, CeFi, DeFi, and TradFi each built out their own institutional infrastructure, though largely along separate product tracks. Exchanges developed institutional custody and prime brokerage services; DeFi protocols introduced professional risk managers, Vault structures, and stricter collateral controls; and traditional financial institutions expanded the tokenization of funds, Treasuries, and money market funds. Innovation during this period remained predominantly isolated within each ecosystem, with limited cross-system integration.

By 2026, the more pronounced change was that these previously disconnected components were directly assembled into the same institutional product architecture: traditional financial assets entered crypto’s collateral system; qualified custodians became the entry point for institutions into DeFi; and Risk Curators, Vault infrastructure, and DeFi protocols were further encapsulated into backend modules, with exchanges, custodians, and prime brokers handling client relationships, compliance, and product distribution at the frontend.

Consequently, the focus of this article has shifted from whether institutions are adopting crypto assets to a more specific question: whether the overall distribution, custody, collateral, and risk architecture for the institutional use of digital assets has entered a new phase of restructuring.

Why Now: Three Driving Forces

This modular architecture was not designed by any single institution. It is the result of three forces acting simultaneously.

Driver 1: Since 2022, institutions have become hesitant to leave capital on exchanges.  Following the collapse of FTX in 2022, “keeping assets on an exchange” shifted from standard operating practice to something that required explanation to a risk committee. This change in attitude is visible in onchain data. According to CryptoQuant, ERC-20 stablecoin reserves held on centralized exchanges fell from a peak of more than $75 billion at the end of 2025 to $61.8 billion in July 2026, a decline of about 17.6%. By mid-August, reserves were around $64 billion, roughly 20% below the peak (Figure 2, left). Over the same period, total stablecoin supply fell only from roughly $315 billion in May to $300.86 billion in August, a decline of about 4.5%. The gap between those figures suggests that capital is leaving exchange accounts, but not leaving the crypto market. CryptoQuant also noted that Binance’s share of the remaining reserves rose from just above 60% at the end of 2025 to 68.5%, indicating that as liquidity shrinks, it is also concentrating among the leading venues.

Driver 2: The interest cost of idle margin became unignorable.  Trading on centralized exchanges requires institutions to pre-fund accounts with margin capital, which yields zero returns while locked up. Based on the $61.8 billion exchange stablecoin reserve baseline and a 3-month U.S. Treasury yield of 3.86% (as of Sept. 1, 2026), the crypto industry foregoes roughly $2.39 billion in annualized interest income (Figure 2, right). Before 2021, when short-term U.S. interest rates hovered near zero, this opportunity cost was negligible. However, the aggressive rate hikes post-2022 forced institutions to confront the carrying cost of every dollar in idle margin. That figure is both the value ceiling that yield-bearing collateral can unlock and the soft spot of the entire architecture: it rests on the premise that short-term rates remain high enough.

Driver 3: On the supply side, all three parties have readied their modules.  Demand side alone is not enough; the supply side also needs the capacity to absorb it. As of July 2026, onchain Real World Assets (RWAs)—excluding stablecoins—reached approximately $33.5 billion (rwa.xyz), with tokenized U.S. Treasuries rising from about $13.4 billion in early April to nearly $15 billion by May, giving TradFi assets that can serve as collateral. Morpho and Aave completed their shift in 2025 from “applications for DeFi users” to “backend infrastructure for institutions.” Meanwhile, the EU’s MiCA framework took full effect on July 1, 2026, and BitGo’s OCC-chartered national trust bank entity began custodying DeFi position tokens, giving “licensed custodian” a clear regulatory meaning.

II. TradFi Collateral Begins Entering Crypto Market Structure

From “Moving Assets Onchain for Trading” to “Keeping Assets in Custody as Margin”

The partnership between Franklin Templeton and Binance stands out as the most representative case of 2026. The two firms announced a strategic collaboration in September 2025, and their institutional off-exchange collateral program officially launched in February 2026. Under this setup, eligible clients can post tokenized money market fund shares issued via Franklin Templeton’s Benji technology platform (BENJI, representing the Franklin OnChain U.S. Government Money Fund, ticker: FOBXX) as off-exchange collateral for trading on Binance. The value of these fund shares is “mirrored” within Binance’s trading environment, while the underlying tokenized assets themselves are held off-exchange by Ceffu as the custody layer.

The significance of this structure lies in how the tokenization of traditional financial assets is moving away from the past emphasis on “moving assets onchain for issuance and toward more practical collateral and balance-sheet management scenarios. Institutions can continue to hold tokenized money market funds within a regulated custody structure while using those assets as margin for crypto trading, without first having to convert the assets entirely into stablecoins and transfer them to an exchange account.

It also addresses two longstanding concerns for traditional institutions. First, it mitigates counterparty risk. By keeping assets in regulated custody off-exchange while the exchange merely records a mirrored value on its risk ledger, institutions no longer have to expose their full principal to the credit risk of a single exchange. Second, it enhances capital efficiency. Pledged assets remain invested in underlying U.S. Treasury bills or money market funds, continuously generating interest as yield-bearing collateral. As a result, the value of tokenized assets is evolving from a pure investment product into a programmable balance-sheet tool that can move across financial scenarios and be utilized repeatedly.

Breaking the Structure Down: Who Does What?

Mapping this setup against the five functions outlined in Section 1 reveals a clear division of responsibilities:

●      Asset ownership: The fund shares remain in the custody structure throughout, and the institution retains legal ownership. FOBXX is the first U.S.-registered mutual fund to use a public blockchain as its official system of record. Ownership of the shares is governed by the existing legal framework for funds under the Investment Company Act of 1940, rather than by custom smart-contract logic.

●      Custody: Ceffu holds the tokenized shares; the exchange does not touch the underlying assets.

●      Valuation: The fund is valued based on Net Asset Value (NAV) published daily by the fund manager. The exchange then sets the collateral haircut on that basis.

●      Execution: Positions are executed on Binance’s matching engine.

●      Liquidation: Margin call triggers and liquidation procedures are set by the exchange’s risk-management system, without involving token-governance voting. However, those parameters are not public, making it impossible for outside observers to assess how conservative they are.

●      Loss and Liability: The revenue-sharing and liability arrangements among the parties have not been disclosed and cannot currently be assessed.

Why This Direction Demands Attention

If tokenized U.S. Treasuries, money market funds, and other high-quality assets become broadly accepted by CEXs, OTC desks, prime brokers, and DeFi lending markets, then one of the largest future markets for RWAs might not be secondary market trading, but rather serving as a more foundational collateral layer for the broader digital asset financial system.

The data supports this view. Turnover rates for tokenized Treasuries and money market funds have remained consistently low, with most supply circulating via primary mints and redemptions rather than active trading. For instance, BENJI held an onchain market capitalization of 687million across 1,121addresses(asofSept.9,2026),yet recorded a monthly transfer volume of only18.17 million (via rwa.xyz). If the primary use of these assets is to serve as collateral rather than trading instruments, the benchmark for success should shift to collateral acceptance—that is, how many venues are willing to accept them as margin.

Three Unresolved Issues

The share of assets actually pledged as collateral is unknown.  No issuers currently disclose what percentage of their tokenized money market fund AUM is actively pledged as margin collateral versus passively held for yield generation. BENJI’s monthly transfer volume—under $20 million—at least suggests that the vast majority of shares remain in a “buy and hold” state. Existing analysis notes that institutional holders of tokenized Treasuries tend to prefer passive holding, remaining unwilling to pledge assets into positions exposed to smart contract risk.

Redemption timing is mismatched.  Money market fund subscriptions and redemptions follow traditional settlement cycles, whereas crypto market liquidations run 24/7 in real time. During periods of extreme market volatility that trigger margin calls, whether the collateral can be converted quickly enough to keep pace with the liquidation engine requirements remains unverified under actual stress events.

The model depends on interest rates.  If the 3-month U.S. Treasury yield declines from its current 3.86% to 2.00%, the industry-wide annualized yield loss shrinks from $2.39 billion to $1.24 billion, and the advantage of yield-bearing collateral over directly using stablecoins narrows accordingly—while the fixed costs of custody, mirroring, and curation do not decline proportionately.

III. CeFi Frontend + DeFi Backend: A Shifting Distribution Architecture

Why Institutions Could Not Access DeFi Before

Another more pronounced change in 2026 is the reversal of how DeFi distributes to institutions. Previously, gaining lending or yield exposure required institutions to set up their own wallets, manage signing permissions, choose protocols and markets, and handle onboarding and risk management across different chains and protocols on their own. Even for relatively mature protocols, institutional adoption remained constrained by operational complexity, compliance requirements, and internal risk approvals. A compliance department at a traditional asset manager could rarely approve “transferring funds into an anonymous smart contract”—even one that had operated securely for five years.

2026: DeFi Proactively Moves into Institutions’ Existing Accounts

During the first half of 2026, leading custody platforms almost simultaneously embedded Morpho, Aave, or managed DeFi vaults into the accounts and custody workflows that institutions already use:

●  Fireblocks launched Earn on April 15, embedding Aave and Morpho directly into its platform serving over 2,400 institutional clients. Sentora was the first curator, with Galaxy joining as a curator on July 16.

●  BitGo opened access to Aave, Spark, and Tesseract via the Narval gateway on June 9, and announced a partnership with Morpho on June 22 to launch an institutional vault, with position tokens held under custody by its OCC-chartered national trust bank entity.

●  Kraken partnered with Upshift on July 15 to launch a customized institutional vault, where positions are held as receipt tokens in clients’ segregated custody accounts, with no pooling and no rehypothecation.

Institutions can continue using their familiar custodians, prime brokers, or treasury systems, while DeFi lending, yield, and liquidity are handled in the backend.

The scale effect of this path is evident in Coinbase’s collateralized lending product. When a user borrows USDC against BTC through the Coinbase interface, the backend converts the BTC into cbBTC and deposits it into Morpho contracts on Base. The product launched in January 2025, surpassed $1 billion in cumulative originations eight months later, reached $2.17 billion on April 14, 2026, and hit $2.3 billion by mid-May. The BTC collateral limit has been raised to $5 million (Figure 3). Max Branzburg, Coinbase’s head of consumer products, described it bluntly to Decrypt: “Coinbase is not lending to users. Coinbase is not participating in the financing itself.” Morpho, for its part, has described the model as “centralized in the front, permissionless in the back.”

This distribution structure can be summarized as follows:

Institutional Client → Existing Institutional Relationships (Kraken / BitGo / Fireblocks / Anchorage / Taurus) → Custody & Governance Layer → Risk Manager / Asset Manager (Galaxy / Sentora / Upshift) → Vault Infrastructure → Lending Protocols (Morpho / Aave) → DeFi Liquidity

What Is Really Changing Is Distribution, and a Longer Chain Reduces Visibility

Between 2023 and 2025, the prevailing logic was that institutions had to actively enter DeFi. Since 2026, the opposite path has  emerged: DeFi is actively moving into the financial accounts institutions already maintain. If this model continues to expand, institutions may eventually not even need to know whether their yield comes from Aave, Morpho, or another lending market—simply allocating an “institutional USD yield strategy” through a familiar custody or treasury interface. This means DeFi protocols themselves may gradually shift from a product facing end users directly to invisible liquidity infrastructure in the backend of financial institutions.

One cost is worth noting here. The lengthening distribution chain reduces end institutions’ visibility into underlying risks, even though the party that bears losses has not changed accordingly. When an institution clicks “Allocate” on a custody interface, its understanding of the risks at each of the five layers below may be no better than when it opened its own wallet in 2024. This issue will be explored in detail in the case study in Section V.

IV. From Embedded DeFi to Managed Embedded DeFi

After Access: Who Decides Where Capital Goes?

Simply integrating Aave or Morpho into a custodian is not the end point of this trend. For institutions, the hard part is not only “how to enter DeFi,” but “once in, who is responsible for choosing markets, setting risk exposure, and adjusting the portfolio in abnormal situations.” That makes the next layer of change in 2026 worth watching closely: the formation of Managed Embedded DeFi. Under this model, custodians or institutional platforms handle client relationships, compliance, and the asset-management entry point; specialized risk managers or curators such as Galaxy, Upshift, Sentora, Steakhouse, and Gauntlet handle strategy design, asset selection, exposure caps, and ongoing risk management; and Vault infrastructure such as Morpho translates those mandates into portfolios that can be executed onchain.

This structure is in effect recreating the financial division of labor familiar from TradFi. Rather than managing each lending pool themselves, institutions choose a professional manager or risk mandate; the custodian handles asset safety and account management; the manager decides portfolio allocation; the Vault provides a standardized, fund-like structure; and the DeFi protocol handles only the most basic liquidity and execution. As this model matures, the choices institutions make may gradually shift from “Should I use Morpho?” to “Should I allocate to Galaxy, Upshift, or a strategy managed by a specialized curator?” That implies risk curators may gradually evolve into a new kind of onchain asset manager.

This new role has already reached considerable scale. According to DefiLlama, the risk-curator sector currently manages about $9.26 billion, of which Steakhouse Financial accounts for about $2.99 billion, or 32.3%, and Gauntlet ranks second at about $1.51 billion—the latter completed a $125 million Series C led by SBI in July 2026. Concentration is high: as of November 2025, the top four curators controlled about 65% of curated capital.

Profit Sharing: How Yield from Collateral Is Distributed

Behind the division of responsibilities lies a contest over profit distribution. The following model is built on publicly disclosed fee ranges to compare the relative bargaining power of each module; it does not represent the actual pricing of any specific transaction.

Assume the collateral is a tokenized money market fund, with the underlying yield based on the 3-month U.S. Treasury rate of 3.86%—that is, a yield pool of 386 basis points per $100 per year. Based on median publicly disclosed industry rates: qualified custody charges about 12 bps, collateral mirroring and off-exchange settlement about 8 bps, Vault infrastructure about 10 bps, lending protocols about 15 bps, and risk curators charge a percentage of yield. Curator performance fees in practice typically fall in the 5% to 15% range (Morpho Vault’s cap is 50%, but few curators approach it); this article takes the 15% upper bound as an assumption, translating to about 58 bps. Intermediary layers total about 100 bps, around one-fourth of the total yield pool (Figure 5). If the curator fee is taken at the 5% lower bound, intermediary layers total about 64 bps, falling to about 16.6%.

This model suggests three things.

Curators have the strongest bargaining power among the intermediary layers, but this position has not yet been fully tested by price competition.  Under the 15% assumption, curators take more than half of intermediary-layer yield, while neither bearing custody responsibility nor making capital payouts for losses. This fee can persist partly because the curator role is still new and institutional clients have not yet developed bargaining habits, and partly because concentration at the top is high. If the curator layer consolidates, fees have room to fall; if a major loss directly caused by curator decisions occurs with no one paying compensation, the pricing basis for this link will be re-examined.

The intermediary share is highly sensitive to interest rates.  Most fees other than curator fees are charged on assets under management. When short-term rates fall, the numerator shrinks while fixed fee rates stay the same, so the intermediary share rises passively. At a 2.00% rate scenario, the same fee structure would occupy about 37.5% of total yield, at which point the economics of institutions building their own channel would become attractive again.

The exchange’s position in this split is not secure.   In an off-exchange collateral structure, the exchange gives up the float income on margin balances in exchange for trading volume and fees. If the same collateral can be freely mirrored across multiple exchanges, the exchange’s ability to lock in client balances declines, and its revenue would depend more purely on execution quality and liquidity depth.

The Playbooks of the Five Parties

The modular architecture has taken shape because the interests of the five parties happen to fit together under the current rate and regulatory environment. The custodian’s core asset is “clients already have accounts here”; they want to keep client assets in their own system and charge one more layer of service fees. The exchange’s core asset is matching depth; they fear balance flight most, so they are willing to give up float income in exchange for not losing trading volume. The asset manager’s core asset is issuance capability and regulatory endorsement; they want their fund shares accepted by as many venues as possible. The protocol side has realized that going directly to institutions does not work, and that retreating to the backend can bring in greater capital. The curator occupies a new position not yet defined by regulation, holding the power to set risk parameters but not bearing custody responsibility. None of the parties is voluntarily giving up margin.

On July 22, 2026, SEC Commissioner Hester Peirce issued a statement titled Headstands and Summervaults, explicitly noting that onchain vaults and lending strategies do not automatically fall outside the scope of securities law simply because they run on smart contracts. The statement did not name any institution, but it targets precisely this gap: “curators are performing asset-manager functions without being subject to an asset-manager liability framework.”

V. Risk Management Evolves from a Protocol Feature into Standalone Financial Infrastructure

Investment Mandates Are Becoming Code

As institutional participation and asset scale expand, risk management in DeFi is evolving from an internal protocol function into a standalone industry layer. Early lending protocols relied primarily on fixed Loan-to-Value (LTV) ratios, liquidation thresholds, and governance-driven parameter adjustments. Today, a growing number of specialized institutions handle collateral onboarding, market simulation, capital allocation, exposure caps, real-time monitoring, and emergency response—increasingly writing these rules directly into Vaults and smart contracts to convert traditional Investment Mandates into executable “Investment Mandates as Code.”

The significance of this change lies in the unbundling of functions traditionally concentrated within a single financial institution. Portfolio managers, risk committees, custodians, security teams, and compliance departments are being split into different specialized service providers across the digital asset ecosystem: curators determine assets and market exposure, risk engines dynamically adjust collateral parameters, security firms monitor protocol and wallet anomalies, and custodians control final asset access. DeFi is not simply “eliminating financial intermediaries”; rather, it is further subdividing intermediary functions and letting each type of risk be borne by more specialized infrastructure.

This Layer Has Not Yet Been Adequately Stress-Tested

A theoretical division of labor can only be judged in real loss events. Two incidents between November 2025 and April 2026 provide two different transmission samples and expose the current weaknesses of this emerging industry layer.

Stream Finance: How a Single Loss Was Amplified Threefold.   Stream Finance is a protocol issuing the yield-bearing stablecoin xUSD. On November 4, 2025, it disclosed that an external fund manager had caused about $93 million in losses, and immediately suspended deposits and withdrawals. Within 24 hours, xUSD fell about 77%, and about $160 million in deposits were frozen. This was originally a single-point loss, but it ultimately evolved into about $285 million in cross-protocol debt exposure, an amplification of about 3.1x (Figure 4, left). The mechanism of amplification was looping: xUSD was taken to lending markets such as Morpho, Euler, Silo, and Gearbox as collateral to borrow real USDC, then used to buy more xUSD, then pledged again. Days before the incident, onchain analyst Cbb0fe noted that Stream’s actual backing assets were only about $170 million, while total borrowing reached $530 million. The damage spread outward along the collateral chain: Elixir’s deUSD, which had lent about 65% of its reserves to Stream, fell about 98% and was ultimately wound down; curator TelosC’s exposure was about $124 million, MEV Capital about $25 million, and Re7 Labs about $14.65 million. Within a week, about $1 billion was withdrawn from DeFi yield products.

Post-mortem analyses revealed that multiple lending markets had hardcoded xUSD’s oracle price at $1. The original intent was to avoid triggering cascading liquidations during minor peg deviations, but the result was that no liquidations were triggered at all during a true depeg: borrowers walked away with real USDC, while lenders were left holding now-worthless collateral. According to a post-incident statement by Re7 Labs, its due diligence team had noticed xUSD’s centralized counterparty risk, but still opened the market based on user demand. “Identifying the risk yet choosing to open the market” is a typical case of the mismatch between fees and responsibility.

KelpDAO: Aave Made No Mistake Yet Bore the Loss.  On April 18, 2026, KelpDAO’s cross-chain bridge was exploited. The attacker leveraged a vulnerability in LayerZero’s message verification to mint about 116,500 rsETH—worth about $292 million, or about 18% of circulating supply—without any real ETH being locked. Rather than selling these tokens, the attacker deposited them into Aave V3 and V4 as collateral to borrow real WETH. Because the Chainlink oracle had not yet reflected the depeg, Aave priced these unbacked assets at pre-attack levels. Aave’s incident report estimated that about $190 million in loans corresponded to unbacked collateral, with final bad debt between $123 million and $230 million.

This time, Aave’s own contracts had no vulnerability, but the market reaction was “run first to be safe”: WETH depositors, fearing that those who withdrew later would bear the residual losses, withdrew collectively. On the first day, Aave’s TVL fell by about $6.6 billion; within 48 hours, $8.45 billion in deposits flowed out, and DeFi’s total TVL shrank by $13.21 billion over the same period (Figure 4, right). Aave’s Umbrella security module held only about $50 million at the time, and in the end, Lido, EtherFi, and Aave founder Stani Kulechov each proposed injecting 5,000 ETH to fill the gap. SparkLend, which had reduced its rsETH exposure beforehand, absorbed $1.4 billion to $1.7 billion in new deposits within days.

Common Features of Both Cases

Putting the two incidents together, three features of this emerging risk infrastructure layer can be identified.

First, isolation design limited direct losses, but it did not cover the oracle-pricing link, nor did it limit the spread of confidence shocks across protocols. Second, the path of risk transmission has changed, and it is faster. In traditional finance, losses at one fund must pass through clearinghouses, settlement cycles, and windows for regulatory intervention before reaching another institution; onchain, these buffers do not exist, and a single erroneous oracle price can transmit losses to three or four protocols within hours. Third, the mismatch between fees and responsibility appeared in both cases: in the Stream incident, curators charging performance fees identified the risk yet still opened the market; in the KelpDAO incident, the losses were borne by depositors and by protocol parties that voluntarily contributed capital, while the cross-chain bridge that caused the flaw did not have equivalent capacity to compensate.

VI. The End State: A Unified Institutional Digital-Asset Stack

One Stack, Not Three Separate Systems

If the above trends continue, the institutional digital-asset market may no longer be divided into three separate systems—CeFi, DeFi, and TradFi—but may gradually form a unified institutional stack. At the base layer would sit tokenized Treasuries, money market funds, stablecoins, BTC, ETH, and other assets. These assets are held through qualified custodians or the banking system, then enter a unified collateral-management layer, and are deployed into CEXs, OTC, DeFi lending, or Vault strategies depending on different needs. Risk curators, oracles, security monitoring, compliance, and reporting provide control throughout the process.

The resulting stack could look like this:

Tokenized assets / stablecoins / crypto assets → qualified custody → collateral management → CEX / OTC / DeFi execution → vaults / DeFi liquidity → risk curators / risk engines → compliance / monitoring / reporting

The Key Question: Who Owns Which of the Six Roles?

Under this architecture, the key question in the future is no longer whether a service belongs to CeFi or DeFi, but where the assets are, who holds control, who is responsible for valuation, who provides liquidity, who sets risk parameters, who bears default and operational risk, and whether these different components can be managed under a single institutional balance sheet. The table below consolidates the content of the first five sections.

Dimension

Platform-Based Finance (exchange-led model)

Modular Institutional Finance

What Remains Unclear

Asset Ownership

Assets transferred to an exchange typically become liabilities of the exchange, and clients hold a claim.

Assets remain with licensed custodians; clients retain legal ownership. The exchange records only a mirrored value.

Who prevails when the custodian and the exchange disagree on reconciliation; the priority of claims on collateral if the custodian defaults.

Custody

Exchange-operated wallets.

Independent licensed custodians (Ceffu, BitGo, Fireblocks, etc.).

Position receipts sit with the custodian, while the underlying assets have entered protocol contracts—who bears contract-level losses first.

Valuation

The exchange calculates based on its own index prices.

Tokenized funds are valued at NAV daily; crypto collateral is quoted by oracles.

NAV updates once a day while the liquidation engine runs every second—how to handle the time gap in extreme markets; the risk of hardcoded oracle prices.

Liquidity

Provided directly by the exchange’s internal order book.

Order execution takes place on-exchange, while collateral realization relies on primary fund redemptions or DeFi markets.

Multi-day money market fund settlement cycles fail to align with 24/7 automated liquidation engines.

Liquidation

Set and executed by the exchange’s risk department.

The exchange sets margin parameters; on the DeFi side, curators set LTVs and thresholds.

Exchange haircuts and trigger conditions are mostly undisclosed; curator parameter adjustments themselves may become liquidation triggers.

Loss Liability

Borne by the exchange, or its insurance fund.

Dispersed across protocol security modules, curator reputation, custodian insurance, and frontend contractual liability.

Curators charge performance fees but make no capital payouts for losses; custodians hold the receipts but are not liable for protocol losses; ultimately the loss often falls on depositors.

The final row is the most noteworthy. In the platform model, loss responsibility is clear: clients go to the exchange, the exchange has an insurance fund, and if all else fails, there is litigation. In the modular model, every party has a plausible reason to say, “This is not my responsibility.” The protocol says the contract had no vulnerability. The curator says the parameters were set according to the framework. The custodian says it is only responsible for safekeeping the receipt. The frontend says the risk was already disclosed at the protocol level.

Short-Term Integration, or Long-Term Structural Change

This brings us back to the question at the start of this article: are the cases that appeared in 2026 merely short-term product integrations, or do they point to a longer-term structural change in the market—that is, digital asset finance shifting from platform-based finance, which relies on a single platform for custody, execution, credit, and yield, to modular institutional finance assembled from multiple specialized modules.

Our judgment is conditional. The change on the demand side is structural.  The continued contraction of exchange stablecoin reserves shows that institutions are systematically reducing exchange balances, and this trend will not automatically reverse with a market recovery. The capacity on the supply side is already in place.  As of the first half of 2026, the four key modules—tokenized Treasuries, licensed custody, encapsulable lending protocols, and curators—already had live products in operation. But whether this architecture can be sustained depends on two conditions that have not yet been met.  First, short-term rates must remain at a level that covers intermediary costs; a rapid decline in rates would loosen the economic foundation of modularity before any regulatory or competitive change does. Second, a framework for loss responsibility must be clarified, whether through regulation (the SEC statement is one signal) or through a sufficiently large loss event that forces the market to reprice curators.

From here, three evolutionary paths may emerge over the next one to two years. They are not mutually exclusive, but they would produce different winners.  Path 1: Frontend integration of curation.  Custodians and exchanges realize that curators are the most profitable and least liable link in the value chain, so they bring curation in-house through internal buildouts or acquisitions, and the industry returns to a two-layer structure of “large frontend + backend protocols.” Path 2: Curators are brought under regulation.  Onchain Vaults are recognized as investment products requiring registration, curators assume fiduciary duties similar to those of asset managers, fees move closer to traditional asset management, and the barrier for institutional capital to enter falls accordingly. Path 3: An interoperable collateral network takes shape.  The same tokenized fund share can be freely mirrored across multiple exchanges, pushing exchanges back to a pure execution layer, with competition returning to matching depth, latency, and fees.

The following three indicators can be observed over the next two quarters to test the judgment of this article: whether curator AUM recovers after the next major loss event; whether the same tokenized collateral is accepted by more than two trading venues at the same time; and whether new institutional clients in off-exchange collateral programs slow if the 3-month Treasury yield falls below 3.00%.

The gains of modular architecture in capital efficiency can be calculated; its effectiveness in risk pricing has not yet been proven. The restructuring of the architecture changes the path of risk transmission and the party that bears losses, but the total amount of risk has not shrunk. Understanding this is the starting point for judging how far this round of convergence in 2026 will go.

---

Data in this report is current as of Sept. 9, 2026. All figures may change with market conditions. Figure 5 is an illustrative model based on publicly available fee ranges and does not reflect actual pricing. There is currently no public disclosure of the proportion of tokenized money market funds that is actively used as collateral; the related analysis in this report is based on stated assumptions. This report does not constitute investment advice and does not make any prediction regarding the market performance of any asset.

About HTX Ventures

HTX Ventures, the global investment division of HTX, integrates investment, incubation, and research to identify the best and brightest teams worldwide. With more than decade-long history as an industry pioneer, HTX Ventures excels at identifying cutting-edge technologies and emerging business models within the sector. To foster growth within the blockchain ecosystem, we provide comprehensive support to projects, including financing, resources, and strategic advice.

HTX Ventures currently backs over 300 projects spanning multiple blockchain sectors, with select high-quality initiatives already trading on the HTX exchange. Furthermore, as one of the most active FOF (Fund of Funds) funds, HTX Ventures invests in 30 top global funds and collaborates with leading blockchain funds such as Polychain, Dragonfly, Bankless, Gitcoin, Figment, Nomad, Animoca, and Hack VC to jointly build a blockchain ecosystem. Visit us here.

Feel free to contact us for investment and collaboration at [email protected]

References

Market data and Interest Rates

1. CryptoQuant, “All Stablecoins (ERC20): Exchange Reserve”, https://cryptoquant.com/asset/stablecoin/chart/exchange-flows/exchange-reserve
2. Bitget News, “CryptoQuant says stablecoin exchange reserves fall to $61.8 billion, liquidity risks persist,” July 23, 2026, https://www.bitget.com/news/detail/12560605535968
3. DailyCoin, “Stablecoin Liquidity Falls as Reserves Concentrate on Binance,” Aug. 2026, https://dailycoin.com/stablecoin-liquidity-falls-as-reserves-concentrate-on-binance
4. FRED (Federal Reserve Bank of St. Louis), “Market Yield on U.S. Treasury Securities at 3-Month Constant Maturity (DGS3MO),” https://fred.stlouisfed.org/series/DGS3MO
5. Federal Reserve H.15 Selected Interest Rates, Sept. 9, 2026, https://www.federalreserve.gov/releases/h15/

Tokenized Assets and RWAs

6. rwa.xyz, “U.S. Treasury Funds”, https://app.rwa.xyz/treasuries
7. rwa.xyz, “BENJI,” snapshot as of Sept. 9, 2026, https://app.rwa.xyz/assets/BENJI
8. Stobox, “The State of RWA Tokenization – 2026 Mid-Year Report”, July 10, 2026, https://www.stobox.io/reports/state-of-rwa-2026
9. InvestaX, “Q1 2026 Real World Asset Tokenization Market Report,” April 15, 2026, https://investax.io/blog/q1-2026-real-world-asset-tokenization-market-report
10. Franklin Templeton, Franklin OnChain U.S. Government Money Fund (FOBXX) Prospectus, Aug. 1, 2026, SEC EDGAR, https://www.sec.gov/Archives/edgar/data/0001786958/000165558926000970/c485bpos.htm

Franklin Templeton × Binance Off-Exchange Collateral Program

11. Zawya / joint press release, “Franklin Templeton and Binance advance strategic collaboration with institutional off-exchange collateral program,” Feb. 2026, https://www.zawya.com/en/press-release/companies-news/franklin-templeton-and-binance-advance-strategic-collaboration-with-institutional-off-exchange-collateral-program-sbvvqcb9
12. The Defiant, “Franklin Templeton and Binance Launch Tokenized Collateral Program,” Feb. 11, 2026, https://thedefiant.io/news/cefi/franklin-templeton-and-binance-launch-tokenized-collateral-program
13. Cointelegraph, “Franklin Templeton to Let Tokenized Money Funds Back Binance Trades,” Feb. 2026, https://cointelegraph.com/news/franklin-templeton-binance-tokenized-mmf-collateral

Custodian Integration with DeFi

14. Fireblocks, “Fireblocks Launches Earn, Giving Institutions Native Access to Onchain Lending,” PR Newswire, April 15, 2026, https://www.prnewswire.com/news-releases/fireblocks-launches-earn-giving-institutions-native-access-to-onchain-lending-302742386.html
15. Fireblocks Blog, “Earn on Stablecoin Balances: Fireblocks Launches Native Yield Offerings,” https://www.fireblocks.com/blog/earn-on-stablecoin-balances
16. CryptoDaily, “Galaxy Curator brings Morpho yields to Fireblocks Earn,” July 17, 2026, https://cryptodaily.co.uk/2026/07/galaxy-curator-opens-morpho-yield-fireblocks-clients
17. BitGo, “BitGo Launches Institutional DeFi Access to Aave, Spark, and Tesseract Through Narval Integration,” June 9, 2026, https://investors.bitgo.com/news/news-details/2026/BitGo-Launches-Institutional-DeFi-Access-to-Aave-Spark-and-Tesseract-Through-Narval-Integration/default.aspx
18. Business Wire, “BitGo Expands Institutional Access to DeFi Vault Strategies With Morpho,” June 22, 2026, https://www.businesswire.com/news/home/20260622528152/en/BitGo-Expands-Institutional-Access-to-DeFi-Vault-Strategies-With-Morpho
19. Kraken Blog, “Kraken Institutional partners with Upshift to bring customers custom institutional vaults,” July 15, 2026, https://blog.kraken.com/product/kraken-institutional/upshift-partnership
20. The Block, “Kraken Institutional taps Upshift to build vaults that earn yield on idle bitcoin, ETH and stablecoins,” July 15, 2026, https://www.theblock.co/post/408448/kraken-institutional-taps-upshift-to-build-vaults-that-earn-yield-on-idle-bitcoin-eth-and-stablecoins
21. Kraken Blog, “MiCA enforcement begins July 1: what it means for institutional counterparties,” June 23, 2026, https://blog.kraken.com/product/kraken-institutional/mica-enforcement-begins-july-1

Coinbase × Morpho

22. The Block, “Coinbase tops $1 billion in bitcoin-backed onchain loans via Morpho,” Oct. 1, 2025, https://www.theblock.co/post/373032/coinbase-tops-1-billion-in-bitcoin-backed-onchain-loans-via-morpho
23. Decrypt (republished by Yahoo Finance), “How Coinbase Profits on Bitcoin-Backed Loans as a ‘Technology Provider’,” Oct. 3, 2025, https://finance.yahoo.com/news/coinbase-profits-bitcoin-backed-loans-183845343.html
24. crypto.news, “Coinbase brings $5M crypto-backed loans to UK via Morpho on Base,” April 20, 2026, https://crypto.news/coinbase-brings-5m-crypto-backed-loans-to-uk-via-morpho-on-base/
25. Crypto Briefing, “Coinbase offers crypto-backed loans on staked ETH and SOL, hitting $2.3 billion in originations,” June 16, 2026, https://cryptobriefing.com/coinbase-crypto-loans-staked-eth-sol/
26. Coinbase Help Center, “Crypto-Backed Loans,” https://coinbase-consumer.sjv.io/anKdgR
27. Morpho, “Morpho 2026,” Jan. 16, 2026, https://morpho.org/blog/morpho-2026/

Risk Curators

28. DefiLlama, “Steakhouse Financial – TVL, Fees & Revenue” (including Risk Curators category aggregation), https://defillama.com/protocol/steakhouse-financial
29. DefiLlama, “Gauntlet – TVL, Fees & Revenue,” https://defillama.com/protocol/gauntlet
30. The Big Whale, “Onchain vaults enter the securities perimeter: the coming reckoning for curators,” July 24, 2026, https://www.thebigwhale.io/article/onchain-vaults-enter-the-securities-perimeter-the-coming-reckoning-for-curators
31. Chorus One, “DeFi Curators in 2025: Navigating Chaos, Building Resilience,” https://chorus.one/reports-research/defi-curators-in-2025-navigating-chaos-building-resilience
32. Steakhouse Financial, “DeFi Markets Update 2026-03-17,” https://kitchen.steakhouse.financial/p/defi-markets-update-2026-03-17

Stream Finance Incident

33. CoinMarketCap Academy, “Stream Finance Stablecoin xUSD Crashes 77% After $93M Loss,” Nov. 4, 2025, https://coinmarketcap.com/academy/article/stream-finance-stablecoin-xusd-crashes-77percent-after-dollar93m-loss
34. The Block, “Elixir sunsets deUSD synthetic stablecoin following Stream Finance unwinding,” Nov. 6, 2025, https://www.theblock.co/post/377961/elixir-sunsets-deusd-synthetic-stablecoin-following-stream-finance-unwinding-aims-full-redemptions
35. Tiger Research, “Collapse of the DeFi Jenga: The Stream Finance Breakdown,” Nov. 14, 2025, https://reports.tiger-research.com/p/collapse-of-the-defi-jenga-the-stream-eng
36. HTX Insights, “$93 Million Casually Misappropriated? The Truth Behind the Stream Finance Collapse,” Dec. 16, 2025, https://www.htx.com/news/Project%20Updates-4NfKiUET/?invite_code=9cqt3
37. The Defiant, “Stream Finance Starts Collecting Creditor Claims in Step Toward ‘Global Resolution,’” June 29, 2026, https://thedefiant.io/news/defi/stream-finance-starts-collecting-creditor-claims-in-step-toward-global-resolution
38. Pharos, “Stream Finance: loss broke three stablecoins,” July 2, 2026, https://pharos.watch/learn/case-studies/stream-elixir-contagion-2025/

KelpDAO Incident

39. CoinDesk, “Aave records $6 billion TVL drop as Kelp hack exposes structural risk at DeFi lender,” April 19, 2026, https://www.coindesk.com/tech/2026/04/19/aave-records-usd6-billion-tvl-drop-as-kelp-hack-exposes-structural-risk-at-defi-lender
40. CoinDesk, “The $13 billion DeFi wipeout in two days, and it started with KelpDAO attack,” April 20, 2026, https://www.coindesk.com/markets/2026/04/20/defi-tvl-drops-more-than-usd13-billion-in-two-days-following-kelp-dao-hack
41. CoinDesk, “Aave could face up to $230m in losses after Kelp DAO bridge exploit triggers DeFi chaos,” April 20, 2026, https://www.coindesk.com/tech/2026/04/20/aave-could-face-up-to-usd230-million-in-losses-after-kelp-dao-bridge-exploit-triggers-defi-chaos
42. CoinDesk, “KelpDAO hack news: Aave leads DeFi bailout push after $292M crypto exploit,” April 23, 2026, https://www.coindesk.com/business/2026/04/23/aave-rallies-defi-partners-to-contain-fallout-from-usd292-million-kelpdao-hack
43. The Defiant, “Kelp DAO Loses $293M in Bridge Exploit, Leaving Aave With Over $200M in Bad Debt,” April 19, 2026, https://thedefiant.io/news/defi/aave-price-crash-kelpdao-exploit-whale-dump-rxi8o9
44. Crypto Briefing, “Aave faces $195M in bad debt after KelpDAO bridge exploit as Spark absorbs billions in fleeing capital,” 2026, https://cryptobriefing.com/aave-195m-bad-debt-kelpdao-exploit-spark/

Regulation

45. SG-FORGE, “Societe Generale-FORGE deploys its euro and dollar stablecoins in decentralized finance via its partners,” September 30, 2025, https://www.sgforge.com/sgf-deploys-eurcv-usdcv-in-dex/
46. The Defiant, “Fireblocks Launches Stablecoin Yield Product via Aave, Morpho” (including Aave Arc context), April 15, 2026, https://thedefiant.io/news/defi/fireblocks-launches-stablecoin-yield-via-aave-morpho

The post first appeared on HTX Square.