How the IRIS model works
The process starts with a loan intent. A borrower specifies the asset pair, the amount, the loan duration, and the maximum fixed APY they are willing to pay.
That request goes into an off-chain request-for-quote, or RFQ, system. RFQ is a familiar setup in traditional trading: instead of hitting a public order book, you ask a group of market makers to send you a price directly.
Solvers then return signed fixed-rate quotes. Each quote is backed by a bond, which gives the solver something to lose if it fails to honor its commitment.
Once a borrower accepts, they get their fixed terms. The solver takes on the job of managing the underlying floating-rate position on SparkLend. If variable rates rise during the loan, that is the solver’s problem, not the borrower’s.
No pools, no new liquidity silo
There are no pre-built fixed-rate liquidity pools sitting inside the protocol. Capital comes directly from the underlying venue and from solvers.
The protocol also builds in flexibility for solvers. They can refinance positions during the loan term, including in situations involving overdue rates or grace periods after the loan reaches maturity.
SparkLend’s numbers give IRIS a deep starting pool
SparkLend held a total value locked of approximately $4.66 billion as of November 9, 2026.
Outstanding borrows on the platform stood at about $2.64 billion. That figure reflects 47% growth over the preceding three months.
IRIS describes itself as an origination layer that sits atop variable-rate lending venues, “starting with SparkLend.” The phrasing leaves room for expansion to other markets, though no additional integrations are part of this announcement.
What this means for borrowers and solvers
For borrowers, the immediate benefit is predictability. A user who knows their cost of capital for the next 30 or 90 days can plan positions with far less guesswork than someone watching a variable APY drift.
The maximum-APY parameter is a useful safeguard here. Borrowers set a ceiling, and if no solver can beat it, no deal happens.
For solvers, IRIS creates a new business line. They earn the spread between the fixed rate they quote and the variable rate they actually pay on SparkLend, assuming their rate forecasts hold up. When those forecasts miss, their bonds are on the line.
The off-chain RFQ component also introduces a layer that is not fully onchain. That is common for RFQ systems and helps with speed and pricing, but it does mean users are relying on the quoting infrastructure to function smoothly.
The metrics worth watching are straightforward. How many solvers participate, how competitive their quotes are relative to SparkLend’s floating rates, and how much of SparkLend’s roughly $2.64 billion in outstanding borrows eventually routes through IRIS will show whether fixed-rate borrowing has found its footing on this design.
