There is a large pool of assets that could be earning yield onchain, and there is a growing number of funds and users looking for diversified yield opportunities onchain. No single strategy can fit every balance sheet or customer. That’s why IPOR builds programmable yield infrastructure onchain, supplying DeFi with the operating system for composability.
Fintech, payment providers, CEXs, banks, you name it. Everyone is rushing to build and deliver products onchain. That’s what their customers demand, and why competition is rising so fast. However, building and operating the stack for running yield strategies is a slow and highly specialized process. Fusion by IPOR satisfies that need, letting institutions offer a configurable set of yield products without giving up the client relationship.
Offering generic vaults packaged as Earn products is no longer enough. Different clients want different assets, liquidity terms, strategies, and levels of risk. Building that engine internally means integrations, accounting, automation, monitoring and risk controls, and a team capable of operating it all.
Fusion turns that infrastructure into a programmable onchain engine. With it, anyone—from fintech to CEXs, wallet providers or funds, to AI agents—can build yield products around their own clients and risk parameters, automate how capital moves within predefined limits, and expand the strategy set without rebuilding the stack each time.
After four exploits that took $939M between them, from Stream to Kelp, with the two largest coming 17 days apart in April 2026, deposits in Fusion vaults reached a record $71.6M on Sep 24, after they fell by more than half twice.

Takeaways
- Idle assets have an opportunity cost. Wallets, exchanges, fintechs, treasuries, and asset managers need ways to put client assets to work.
- One yield product cannot serve every client. Different clients need different risk, liquidity, and return profiles; managers should be able to combine protocol integrations and strategy logic with pre-defined vault limits
- Fusion makes yield configurable. Managers choose markets, limits, and strategies, turning one infrastructure stack into many yield products.
- Automation stays inside the vault and hard limits. Operators can execute strategies without gaining withdrawal rights.
- One stack can support many products. Fusion vaults can range from simple lending to leveraged and diversified strategies with RWAs and tokenized stocks.
- You keep the client. Fusion supplies the yield infrastructure; the institution owns the yield product, brand, and distribution.
Build the Product, not the Infrastructure
An Earn button hides an operating business, a particularly complex one. To name a few requisites: integrations and adapters connecting assets to yield venues, audited vault code, real-time accounting, and access checks.
Each step carries dependencies: some are known, others are hidden. For example, a strategy that simply lends a stablecoin on one market depends on that market's oracle and its collateral list, but it also depends on the protocols that feed that market. Usual's USD0++ fell off its peg in January 2025 and unsettled other DeFi markets. Stream disclosed a $93M loss due to opaque accounting. After the KelpDAO exploit, Aave froze its rsETH markets and withdrawals hit protocols that held no rsETH.
IPOR built this desk for itself first, because they are veterans in the space. The team has run onchain rate products since 2022, starting with interest rate swaps. Back then they were pioneers in capital efficiency by taking idle liquidity from the pools that underwrote swaps to dynamically allocate the capital to Aave and Compound.
Next, the IPOR DAO approved the build of Fusion in May 2024, to take their asset management engine to the next level. Most recently, in September 2026, IPOR published a launchpad that builds and rehearses a vault from one strategy file. Tesseract, a manager authorized under the EU's MiCA rules, has been working with Fusion's contracts for more than a year before it made them the main destination for its institutional clients in March 2026.
Institutional adoption moves at that pace: one diligence review and one client at a time, gradually and behind the scenes.
How a Fusion Vault Works
Fusion builds every vault from fuses. A fuse is a small contract that performs one action on one protocol, such as a deposit on Aave or a loan on Morpho. A manager combines fuses into a strategy, from a single-market lending vault for a cautious client to a staking loop for an aggressive one. IPOR's fuses reach more than 40 protocols through 170 methods, and its factory can deploy a new vault in one click.
While being a permissionless protocol, IPOR also designed Fusion for regulated asset managers. The institution picks the fuses and sets the limits, just like the permissionless path, while IPOR supplies the integrations and the accounting. This allows an institution to start with a working vault instead of an engineering plan, frontrunning its competition. As of Sep 26, fourteen vaults hold $1M or more, and together they carry $62.6M of the $71.6M in deposits.

Each vault does one job. A vault's configuration lists the markets it can enter and caps what it can place in each, and any other transaction fails. For instance, Tesseract announced that each of its Dedicated Client Vaults serves one client and keeps that client's assets apart. As a comparison, each vault acts like a separately managed account, which is the structure asset managers already use for single clients.
The strategy and the money stay decoupled. Vault admins set which fuses and markets a vault can use. The Alpha, the operator who trades, acts only through those fuses. The Alpha cannot move client funds out. Institutional vaults can also admit only wallets that passed KYC or KYB checks. Similarly, any change to fees, assets, or fuses waits behind a timelock, allowing depositors to withdraw before the change takes effect.
Anyone can check the books. Every vault prices its shares from the positions it reads onchain, and anyone can recompute those balances to verify. Unlike protocols like Stream Finance, which failed in November 2025 on the opposite design, because its manager priced xUSD from books kept offchain, IPOR reports two numbers for every vault: the deposits and the gross position those deposits carry after borrowing.
The rules also compose. A Fusion vault can hold shares of any vault built to the ERC-4626 token standard, including other Fusion vaults. That way a manager can build a fund of funds, spreading client money across several strategies, each inside its own limits. IPOR's vault data showed $2.97M in 12 such positions on Sep 24, about 4% of vault capital, most of it inside Tesseract's own vaults. Tesseract's private client vaults grew from $0.35M on Mar 1 to $11.3M on Sep 24, 2026.

Around the vault, IPOR has also added parts that an institution usually buys separately. IPOR launched onchain cover for select vaults, provided by OpenCover and underwritten by Nexus Mutual. Blockaid threat monitoring now covers a set of vaults as well. A public, read-only server, launched in July 2026, answers plain-language questions about every Fusion vault, and its launchpad lets an AI agent set up and rehearse a vault with a person in the loop.
Capital Efficiency With Brakes
A dollar in a Fusion vault rarely sits idle. Vaults post collateral on lending markets and borrow against it. They put the borrowed money to work, and one dollar of deposits then supports several dollars of positions. Deposits of $71.6M carried $259M of positions on Sep 24, a multiple of 3.6. The multiple fell from 6.0 at the October 2025 deposit peak to 2.8 by mid-November, briefly topped 6 again around the KelpDAO exploit in April 2026.

One example is the cbETH loop on Base. cbETH is Coinbase's staked ether token, and it earns about 2.75% a year from staking. TAU Labs' vault runs the loop by supplying cbETH to Aave as collateral, borrowing WETH against it, buying more cbETH, and repeating, up to nearly 10 times effective leverage. The spread between staking yield and borrowing cost puts the combined result at about 15% after fees, which has been quite profitable during times of Aave rewards. Fusion positions hold 50.0% of the cbETH on Base, most of it supplied to Aave.

To see the positive-sum game in practice, noticing how lending venues gain from that scale too. Fusion vaults supplied $114M to Aave and borrowed $98.7M from it on Sep 24, while Anchorage's client profile of IPOR stated that Fusion vaults are among the largest sources of borrowing demand on Morpho and SparkLend. Aave carries the largest positions on both sides, with Spark and Morpho next.

Thanks to automation, when Balancer and Stream broke, the vaults cut borrowing fast. The multiple fell from 6.0 on Oct 27 to 2.8 on Nov 13, and gross positions fell 78% from their record while deposits fell 55% by Nov 21, showing how the vaults shed borrowed money faster than client money left. Gross positions fell 78% within weeks of the November 2025 shocks and 62% in spring 2026, then rebuilt to $259M.

Four Shocks and a Record
Four outside shocks hit the vaults between November 2025 and April 2026, and the risk controls held. First, Fusion ran no Balancer code. When Stream's xUSD lost its peg a day later, the DAO's vaults held only collateral that its risk group had approved. In March 2026, none of its vaults had exposure to Resolv. On the day of the KelpDAO exploit, the DAO's stETH vault pulled all of its WETH from Aave's core market, and that the vault kept running without pause. Fusion's own incident came in January 2026, when an attacker took $336K from its oldest vault, which predated the stricter check that later vaults enforce. A white-hat returned the funds minus the 10% bounty and the assets were put back to work within hours. Both declines followed shocks at other protocols, and deposits just passed their 2025 peak on Sep 18, 2026.

Deposits fell by more than half in each of the two drawdowns and came back each time. They closed above their October 2025 peak for the first time on Sep 18, 2026, at $66.09M. Six days later they reached $71.6M as three Reservoir vaults accounted for $25.1M of the $36.2M that left in the first drawdown. One operator's exit drove most of that decline.
New deposits drove the recovery. ETH closed 36% below its price of a year earlier on Sep 23, 2026. In dollars, deposits grew 57%. At constant prices, however, Fusion vaults hold 91% more assets than a year earlier.

Separate Distribution From Strategy
A wallet does not need to become a strategy manager to offer yield, and a strategy manager does not need to build a consumer wallet. Each should do what they do best. A fintech, wallet, or exchange can keep the customer account and brand, as well as its regulatory duties, while a separate operator runs the strategy and sets its vault fees. Fusion supplies the vault infrastructure.
The largest exchanges already work this way. Alea's July report on onchain asset management showed that Coinbase routes in-app USDC deposits into Morpho vaults and that Kraken routes deposits into Veda vaults. IPOR offers the same model to wallets, exchanges, and fintechs, with a dedicated vault for each client.
Just like curators in Morpho, operators can run the strategies and earn the fees under the terms IPOR published for builders. As of Sep 2026, six operators run nearly all of the capital, and no single one runs a quarter of it. One of them is IPOR DAO, whose six vaults held 9% of the capital on Sep 28. Clearstar and Sentinel run nearly half of the capital, and Tesseract ranks third through its private client vaults.

The mix has changed fast. Reservoir ran half of Fusion's capital in October 2025 and almost none of it a year later. Clearstar, Sentinel, and Tesseract together grew from 7% of the capital to 62%.

The DAO takes a thin, fixed cut. IPOR DAO voters approved a DAO share of 0.3% a year of assets and 2% of performance on vaults that third parties run, paid in vault shares. Fusion vaults recorded $0.70M of management and performance fees in the twelve months to Sep 23, 2026, for operators and the DAO combined. Monthly fees peaked at $110K in October 2025 and $116K in January 2026.

Distribution has followed. The Base App and the Coinbase Wallet already feature Fusion, while Anchorage Digital, a US federally chartered crypto bank, gives its clients access to Fusion strategies, and BitGo launched access for eligible clients to allocate to Tesseract's client vaults directly from qualified custody.
The Cost of Waiting
The opportunity grows when platforms gain assets but lack suitable products for them. New chains, new collateral, and more distribution can set one asset manager miles apart from another.
Yield already reaches users through familiar platforms. Coinbase routes in-app USDC deposits into Morpho vaults, while Kraken uses Veda vaults. These arrangements put a consumer interface above externally supplied vault infrastructure. Fusion extends that model to managers who want a vault of their own.

Exchanges and fintechs run their own chains. Kraken launched Ink in December 2024. Robinhood announced Robinhood Chain's mainnet on Jul 1, 2026. Circle launched Arc's mainnet on Sep 16, 2026. Deposits on the three chains closed at $1.73B on Sep 25. Each chain needs a yield product for that money, or its users move their stablecoins to chains that pay them.

New assets became collateral. Coinbase launched tokenized Apple, Nvidia, Alphabet, and Meta shares on Base on Aug 24, 2026. Thanks to its infrastructure allowing for fast integration speed, IPOR launched carry vaults on all four the same day.
The SEC approved tokenized trading on Nasdaq in March 2026. Next it announced a five-year exemption for tokenized-securities venues on Sep 17. Then Aave launched a Base market on Sep 25 that accepts seven Coinbase stocks as collateral for USDC loans.
Once a stock secures a loan, a vault can run a carry trade on it. It deposits the stock, borrows stablecoins against it, and lends them at a higher rate than the loan costs. Fusion already runs this structure on other assets.

Today, MiCA lets EU managers such as Tesseract run client money onchain under a license, while the SEC's exemption gives tokenized-securities venues a five-year runway in the US. A regulated firm now has a legal path to an onchain yield product. What they are missing is reliable, scalable infrastructure to grow their AUM in a safe risk-adjusted manner that lets asset managers customize their own strategies.
Build the Product. Keep the Client
Institutional money arrives slowly, gradually and behind closed doors first, then all at once. Fusion is at the start of that cycle, having already held through four market shocks and a record recovery. Its competitive advantage comes from its design: live accounting, adjustable, and customizable strategies to satisfy the full risk-reward spectrum. That lets a regulated firm hand the strategy to an operator and the contracts to IPOR while they keep their clients and brand.
Disclosures
Alea Research has never had a commercial relationship with IPOR and this report was not paid for or commissioned in any way.
Members of the Alea Research team, including those directly involved in the analysis above, may have positions in the tokens discussed.
This content is provided for educational purposes only and does not constitute financial or investment advice. You should do your own research and only invest what you can afford to lose. Alea Research is a research platform and not an investment or financial advisor.


