USDe loops went live on Solana last week through two venues: Kamino's and Jupiter Lend's new Ethena market, the latter curated by Bitwise. It is USDe's first serious foothold outside of the EVM, and it lands as part of Ethena's push to capture share of Solana's $15 billion stablecoin market. Kamino's market crossed $425M in TVL within 24 hours of launch, the fastest market in its history to clear $400M.

On Kamino/Jupiter Lend, a depositor deposits USDe as collateral, borrows USDG against it, swaps the USDG back to USDe, redeposits, and repeats. The looping vaults run that cycle up to 12.5x in a single transaction. At max leverage, the net spread between USDe's base yield (~4%) and USDG's borrow rate (2.5%) is multiplied by the leverage multiplier.

At 12.5x leverage, you earn USDe's underlying yield on the entire collateral position while paying USDG's borrow rate on the borrowed portion. With USDe yielding 4% and USDG borrowing at 2.5%, that lands at roughly ~21.5% on max leverage.

USDe's base yield is the basis trade Ethena runs onchain: long spot ETH/BTC/Gold, short perps, collect the funding rate. The loop does not create new yield. It multiplies your exposure to that single underlying spread by stacking leverage. At 12.5x, you are running ~12.5x exposure to Ethena's basis spread, minus the cost of borrowing USDG. When perps funding compresses (and it has, repeatedly), the loop's net yield compresses with it. This is the same mechanism that powered every prior USDe loop on Aave and Morpho.

Lending markets typically price collateral through oracles to know when positions cross liquidation thresholds. For volatile assets, that means tracking the real market price minute-to-minute. For pegged stablecoins, real-time pricing creates a different problem: small temporary depegs trigger liquidations the protocol does not actually want to fire. Hardcoding the oracle to the pegged reference asset avoids those false-positive liquidations. The cost is being slow to react when a real depeg arrives.

This pattern is now standard practice across major lending markets. Aave hardcodes USDe to USDT, while Morpho curators have used the same approach for USR, USD0++, and similar assets, with mixed results. The Resolv exploit in March turned hardcoded USR oracles at Morpho into a vehicle for arbitrage: when USR's market price collapsed and the curator-set oracles continued reporting $1, traders bought cheap USR on DEXes and deposited it at face value to drain the vaults. Several Morpho vaults across multiple curators absorbed the damage. The Solana USDe launches inherit both the upside and the downside of that pattern.

Kamino and Jupiter Lend land on the same core design: hardcoded oracles, tight LTVs (12.5x max leverage on both), actively managed borrow curves rather than algorithmic IRMs, and full isolation from the rest of each protocol. The two implementations differ in mechanism and governance.

Kamino writes the assumption directly into the smart contract. The USDe oracle is hardcoded 1:1 to USDT, and if USDe deviates more than ±1% from USDT in actual trading, the market pauses entirely rather than liquidating positions. LTVs are calibrated such that depegs within that ±1% band cannot trigger liquidation at any leverage level. Most EVM implementations of hardcoded stablecoin oracles do not include this pause threshold. They simply let the oracle continue reporting the pegged price during stress, which is exactly the failure mode that propagated bad debt during the Resolv incident. Kamino curates its own parameters and has $2.5B+ in deposits with zero bad debt across three years.

Jupiter Lend gets to the same end through oracle architecture and third-party curation. Bitwise (the $11B crypto asset manager) is the curator, marking the first time a traditional asset manager has been given curation authority on a major Solana lending product. The market is structured as a four-party arrangement: Jupiter Lend provides the market layer, Bitwise sets parameters, Ethena supplies the asset, and Fluid runs the underlying infrastructure. The 92% LTV / 94% liquidation threshold on the USDe → USDG pair only works because the oracle keeps the pair locked 1:1. It does that by pricing both USDe and USDG through Pyth's USDC/USD feed. Neither asset's actual market price affects liquidations. Different mechanism, same trust assumption.

Importantly, both markets are isolated from the rest of Kamino and Jupiter Lend, so a liquidation or oracle failure in another market has no contagion path here.

The Solana launch adds two structural features that gives loopers a degree of confidence about forward conditions that prior USDe campaigns did not offer.

The peg risks are the obvious ones. USDe and USDG, a less-tested stablecoin than USDC or USDT, can both lose their peg. The oracle on both venues is designed to keep USDe priced at $1 regardless of what the market says, and that protection cuts both ways. If USDe's underlying collateral is exploited, stolen, or its basis trade structurally breaks, the secondary market will price USDe accordingly while the oracle continues reporting $1. On Kamino, the market pauses when USDe deviates more than 1% from USDT, which means loopers cannot exit while the pause is active. Lenders who supplied USDG can only be repaid in USDe that the market has already discounted. The hardcoded oracle that protects against small deviations becomes the mechanism that traps everyone during a real one.

USDe on Solana also has bridge risk via LayerZero. The LayerZero OFT bridge that moves USDe to Solana runs on a 4-of-4 DVN configuration, where every verifier must sign for a transaction to clear, with $10M-per-hour OFT rate limits.

A more practical risk when looping is when oracle and market prices diverge in normal conditions. When USDe trades above peg from buying pressure, each loop iteration buys USDe at a market premium the oracle treats as $1, so the spread eats into the position on the way in. The same dynamic runs in reverse on unwind. If USDG trades above USDe when you exit, repaying USDG debt costs more USDe than the oracle implies. The headline 20% APY does not account for these entry and exit costs, and they compound across each loop.