top of page

Taking the temperature on Celsius

Background on Celsius’ position in Aave

Aave’s governance post on May 14th estimated a 15% stETH “depeg” = ~$250m in liquidations, 20% “depeg” = ~$2b. Since this post went live, Gauntlet’s suggestion to increase stETH liquidation threshold to 81% (from 75%) and decrease LTV to 69% (from 73%) has been implemented.

Yesterday, Gauntlet submitted a proposal suggesting again raising stETH liquidation threshold, but to 90% this time, and pause ETH borrowing. The vote is ongoing and currently looks like it won’t pass, but we won’t know until all votes are in. It’s an ambitious move that allows stETH further room to depeg/drop in value before Aave liquidations kick in. Obviously, this poses additional risk to the loaned amount, which in Celsius’ case is a mixture of primarily USDC & DAI.

In addition to a drop in ETH price, the stETH discount/depeg pushes that collateral value even further down. stETH discount currently sits at ~6.6% (1 stETH buys 0.934 ETH).

A reasonable comp for stETH is probably GBTC, considering they both represent a currently unredeemable underlying asset that will hopefully be unlocked at some point in the future. GBTC currently trades at a ~29%+ discount to spot BTC price. If this level of discount/depeg happens to stETH/ETH pair, we’d likely see a significant amount of liquidations across all of DeFi.

For strained & potentially forced sellers like Celsius, this would almost certainly liquidate all Aave stETH positions unless they can continue to find more collateral.

Celsius’ wallet topped up their Aave collateral recently, then pulled in ~US$10m in ETH and USDC about earlier today, likely ready on the sideline to further top up their positions if required.

For now, it looks as though this particular position is relatively healthy - but especially with 3AC also running into problems and dumping stETH, it’s certainly not out of the woods just yet.

Of Celsius’ stablecoin loans on Aave, just shy of ~100m is borrowed in DAI.

MakerDAO exposure to all of this

Following Gauntlet’s proposal, MakerDAO have an ongoing debate about whether to temporarily set Aave D3M to 0, with further discussion at the MOMC meeting in ~2 weeks.

Maker have always been on the conservative side of DeFi, and it makes sense for them to temporarily restrict the D3M module to limit further risk via Aave & Celsius in particular.

According to this Maker forum post, DAI’s D3M module is receiving 1.4% on Celsius’ 100m DAI - even if Celsius were to die, it’s unlikely that 100% of their collateral will become completely worthless. But taking the simplest risk:reward thought process, up to $100m in risk exists and 1.4% doesn’t really feel justified.

Aside from the Celsius risk, Aave’s stETH degen risk also exists, many participants have been cross collateralising with stETH on Aave, but the full extent on unbacked stETH is difficult to quantify.

Personally, I think a more nuanced option would be nice. Rather than turning DC down to 0, maybe a fixed term or an option to scale to 20m or 50m before going to 0 would appear less reflexive.

The purpose of introducing D3M in the first place was to increase utilisation of DAI in DeFi, and a move like this could have some negative PR ramifications, especially with Aave’s influential community. LongForWisdom summarises the steps that need to occur for Maker to take an L on this move well.

All that said — increasing utilisation of DAI is of significantly lower priority than simply surviving. MakerDAO are experts at surviving in DeFi, and these market conditions along with the broader macro backdrop make survival an increasingly difficult task. There are better options than throwing good money at bad right now for 1.4%.

I’ll certainly be keeping an eye on how this develops. Hopefully it doesn’t get too ugly before it gets better, and either way, hopefully the lessons learnt solidify resilience of DeFi infrastructure.

Stay safe out there.



Recent Posts

See All


bottom of page