Stripping out oracles and liquidations doesn't eliminate risk — it just transfers all of it to the lender, who's now effectively writing a put option on the collateral for the loan duration. If the borrower's collateral dumps 80% mid-term, they rationally walk away and the lender absorbs the loss, which means rate-setting on Offerbook has to implicitly price vol and default probability that pool-based protocols handle mechanically. Loopscale already proved order-book lending on Solana can work but also got clipped for ~$5.8M through contract exploits — so Jupiter's "simpler architecture" pitch lives or dies on audit quality, not just UX. The NFT and RWA collateral angle is where this gets genuinely differentiated though, since those assets are basically unlendable on Aave/Kamino-style pools that need continuous price feeds.

Top comment by @Benthic