Aave Wants Native Bitcoin Collateral. The Trust Moved, It Didn’t Disappear.

Babylon's Aave RFC would let borrowers lock BTC in Taproot UTXOs and borrow against it on Ethereum — no wrapping, no bridging, no custodian. But redemption now runs through vault providers, Babylon governance, and a 2-of-3 multisig, and your liquidation pays out in WBTC.
The pitch is that your bitcoin never leaves Bitcoin. That is true. It is also not the same as “nobody holds your coins,” and the difference is three governance layers you have never audited.
The claim as filed, and what survives contact with it
Aave Wants Native Bitcoin Collateral. The Trust Moved, It Didn't Disappear.
Spamguy, CC BY 2.5, via Wikimedia Commons

—

On October 14, 2026, Babylon Labs published an Aave Request for Comment proposing native bitcoin as collateral on Aave V4 through its Trustless Bitcoin Vaults protocol. Two new Aave V4 Spokes would be created: a Babylon Spoke, where native BTC is the sole collateral asset, and a BTC Vault Swap Spoke, which handles liquidation settlement. Borrowers would lock BTC in Taproot UTXOs on the Bitcoin network and borrow stablecoins against it on Ethereum. No wrapping. No bridging. No custodian.

If you read the summary paragraph in most coverage, you would conclude that Bitcoin holders can finally use their coins in DeFi without trusting anyone. And the technical claim underneath that conclusion is real: the BTC does stay on the Bitcoin blockchain. I want to take that seriously, because it is a genuine engineering achievement and it deserves better than the marketing it is getting.

So let us peel the claim apart, one layer at a time, and see what is actually being promised.

Layer One: “Not Wrapped” — This Part Is Real

Let me start with what is actually true, because the criticism lands better when it is aimed precisely.

Today, if you want to use bitcoin as collateral on Aave, you use WBTC. WBTC is a token on Ethereum that represents bitcoin held by a custodian. That custodian is a company. It can be compelled, it can fail, it can be hacked, and it can simply refuse to mint. The entire security model of WBTC is the trustworthiness and solvency of an institution holding real coins in a real vault.

The Babylon design removes that specific dependency, and it does so properly. A depositor locks BTC in a Taproot UTXO on Bitcoin. Redemption is governed by on-chain cryptographic rules, specifically zero-knowledge proofs of loan repayment or other events on Ethereum, with settlement occurring directly to a Bitcoin address the depositor controls. The coins never leave the Bitcoin blockchain. There is no custodian holding them. There is no bridge contract that can be drained.

That is not a small thing. For a constituency whose founding slogan is roughly “not your keys, not your coins,” removing the custodian from the picture is the correct first move, and Babylon has made it.

What was actually removed

The custodian. The bridge. The threshold-signature group that conventionally stands behind wrapped bitcoin. All three of those are real single points of failure, and this design genuinely eliminates them. If you were worried about WBTC specifically — about a company, a vault, or a minting process — this proposal addresses your objection directly.

Layer Two: “Not Custodied” — Here It Gets Complicated

Now the second claim, and this is where I want you to slow down, because the proposal’s own documentation contradicts the marketing.

Aave V4 accepts only ERC-20 tokens as collateral. Native BTC is not an ERC-20 token. So the integration creates a representation token — a transfer-restricted ERC-20 that exists solely within the integration’s smart contracts. It is minted when a BTCVault activates and burned when the vault is withdrawn or liquidated. Transfers are restricted to a fixed allowlist: the Aave V4 Hub, the adapter contract, and per-user position proxy contracts. It cannot be traded, transferred between users, or held outside the integration.

On paper, this is an internal accounting unit, not an asset. Fine. But look at what has to happen for it to exist at all.

Before a BTCVault can activate, three types of actors must acknowledge the locked BTC and pre-sign its transaction graph: the vault provider, an infrastructure operator chosen by the depositor; all application vault keepers designated by Babylon for the Aave integration; and all universal challengers in the protocol. Acknowledgments occur only after Bitcoin confirmation. If any actor fails to acknowledge within a timelocked window, the depositor can reclaim the BTC unilaterally.

Read that list again. Your bitcoin is released back to you if a set of third parties fails to act within a time window. That is a genuinely good safety property — it is the correct design, and it is what makes the scheme self-custodial in the meaningful sense. But notice what it implies: the scheme’s security depends on a group of actors performing their roles on schedule, and on a timelock mechanism that has to be configured correctly. The proposal does not disclose the identity of the vault providers or application vault keepers that Babylon intends to appoint, nor does it specify the timelocked windows for activation or the specific threshold for universal challengers.

So the honest statement is not “nobody holds your coins.” It is: your coins are held by cryptography, which is held by a set of actors, whose identities and parameters are not yet public. That is a much weaker trust assumption than a custodian, and it is a much stronger one than zero.

Aave Wants Native Bitcoin Collateral. The Trust Moved, It Didn't Disappear.
Fixedsun, CC0, via Wikimedia Commons

Layer Three: The Three Governance Layers

This is the part I have not seen covered, and it is the most important structural fact in the proposal.

When you deposit into this system, you are not entering one protocol. You are entering three, each with its own decision-making authority, and your position sits in the seam between them.

LAYER 1 - Aave DAO
  controls: collateral listing, borrow caps, supply caps,
            interest rate strategy, oracle configuration
  decides: whether your asset is listed, how much can be borrowed,
           what rate you pay, and when your position is liquidatable
LAYER 2 - Babylon governance
  controls: TBV protocol itself, vault actor sets, protocol config,
            pause and freeze controls
  decides: who the vault providers are, who the challengers are,
           and whether the whole vault system can be frozen
LAYER 3 - 2-of-3 multisig (Llama Risk, Babylon Labs, Aave Labs)
  controls: BTC Vault Swap Spoke parameters
  decides: the sell discount applied to liquidations,
           and the arbitrageur margin

Three layers. Three sets of decision-makers. Three separate places where a parameter can change underneath your position without your consent.

Put the custodial version next to this one, because the comparison is the honest way to see what changed.

WBTC todayAave V4 + Babylon (proposed)
Who can refuse to redeemOne custodianA vault actor set named by governance
Who sets the termsThe custodian’s contractLayer 1 spoke config, Layer 2 parameters, Layer 3 multisig
Recourse if they say noSue a company; it exists in a jurisdictionPost in a forum thread; there is no counterparty to sue
Can you exit unilaterallyRedeem through the merchant processUnilateral reclaim after a timelock, per TBV’s own safety property
Failure modeInstitutional insolvency or compulsionCollective governance drift across three committees

Read that last row twice. The custodial model concentrated trust in one entity you could name, audit, and haul into court. This proposal distributes trust across actors you cannot currently name, none of whom owe you a legal duty. Distributed is better on some axes and worse on others, and pretending it is strictly better is the marketing, not the engineering.

Now consider what each layer can actually do to you. Layer 1 can change the interest rate strategy, which changes your cost of borrowing. It can change oracle configuration, which changes how your collateral is valued, which can move your health factor without any price movement at all. It can set liquidation parameters, which determines how quickly you get closed out.

Layer 2 can change the vault actor set — meaning the parties whose acknowledgment your vault depends on. It holds pause and freeze controls over the vault system. To be fair, the proposal notes that all TBV governance actions execute through a public timelock with on-chain advance notice, which is the right way to do it and should be credited. But a timelock is a delay, not a veto.

Layer 3 sets the liquidation discount and the arbitrageur margin. That is a direct transfer of value from you to liquidators, decided by a three-party committee.

None of this is scandalous. All of it is normal DeFi governance. But it is the opposite of “trustless,” and the proposal’s own architecture section is more honest about it than the announcement is. Users sit between two separate governance systems, with a multisig adding a third configuration layer on top.

Layer Four: The Liquidation Is Not What You Think

Here is the detail that convinced me this design deserves scrutiny rather than applause.

In a standard Aave market, liquidation is permissionless. Anyone can call it. When your position goes unhealthy, an external liquidator repays part of your debt and seizes collateral at a discount. The more liquidators, the faster and more efficiently your position gets closed, and the less slippage you suffer.

In this design, that is not what happens. Only the adapter contract acts as the liquidation manager. Permissionless external liquidators cannot call the Spoke directly. When a position becomes unhealthy, a liquidator calls the adapter to seize a single BTCVault and receive the representation token. The adapter then routes that token through the BTC Vault Swap Spoke, which swaps it for WBTC drawn from an Aave V4 Hub.

So the liquidator does not receive bitcoin. The liquidator receives WBTC, immediately, on Ethereum. That is a deliberate design choice, and it solves a real problem: redeeming the underlying BTC requires a separate process involving authorized participants and a challenge period that can take multiple days. Separating immediate settlement from eventual redemption is the correct engineering answer.

But look at what it means economically. Your collateral is bitcoin. Your liquidation pays out in WBTC. The wrapped BTC is held as debt against the escrowed BTCVault, accruing interest on the Hub. A permissioned set of arbitrageurs can then purchase individual escrowed BTCVaults by repaying the Hub debt and handle Bitcoin redemption through Babylon’s multi-day claim and challenge window.

Permissioned. Not permissionless. A fixed group of arbitrageurs sits between your liquidated collateral and its eventual redemption, and that group’s margin is set by the Layer 3 multisig. The discount applied to your liquidation is a parameter, not a market outcome.

Bitcoin never leaves Bitcoin. Correct. It also never reaches Bitcoin again without passing through a permissioned arbitrageur whose margin was set by committee.

The seam where risk concentrates

Redeeming actual BTC takes multiple days and runs through a challenge window. Immediate settlement uses WBTC. Those two facts create a window in which your collateral exists in two forms at once, on two chains, with two different liquidity profiles, governed by two different protocols. That seam is where this design is least proven, and it is exactly where a stress event would land.

No real money has touched this

The proposal’s own timeline is candid: public testnet deployment for native BTC borrowing was active by October 2026. Testnet is testnet. Real capital has not been deployed against this design, and the remaining ARFC and AIP phases are where proposals get picked apart line by line. That is the correct process, and it is also a statement about maturity.

Aave Wants Native Bitcoin Collateral. The Trust Moved, It Didn't Disappear.
Wolfgang Weber, CC BY 3.0, via Wikimedia Commons

The Context That Should Temper Everyone

Two facts from the same week that the proposal should be read against.

First, Babylon’s scale. The Aave filing states that Babylon’s Bitcoin staking protocol has accumulated more than 100,000 BTC cumulatively staked since launching in August 2024, peaked at 72,000 BTC in total value locked, and currently holds approximately 40,772 BTC staked — worth roughly $3.3 billion at the time of writing, and about 56.6% of its own peak. That is a real, substantial system, and it lends credibility to the infrastructure.

Second, and less comfortably: on October 1, 2026, a third-party module used to loop positions on Aave V3 was exploited for approximately 114.09 ETH, worth about $307,065. Aave’s pools were reported not to be affected, and the incident establishes no defect in the proposed V4 design. But it is a live reminder that the boundaries between Aave’s Spokes, Babylon’s TBV system, and the configuration multisig are not abstract. They are the places where things go wrong, and where the question of who fixes it becomes urgent.

There is also an honest second-order effect worth noting, because it cuts in Aave’s favor. Aave holds one of its largest token supplies in WBTC, but borrow demand for that WBTC has historically lagged. A lot of WBTC sits in Aave’s pools doing very little. Since liquidators in this design get paid in WBTC and borrowers can take out WBTC, the integration could put more of that idle supply to work. That is a real benefit to the protocol, and it is worth asking whether it is also a reason the incentive to approve runs ahead of the incentive to scrutinize.

What I Would Want Answered Before This Goes Live

I am not opposing this proposal. Removing custodians from bitcoin collateral is the right direction, and the unilateral-reclaim safety property is well designed. But these are the questions I would want answered in the ARFC thread before a single satoshi moved, and they are all answerable from public documents.

  • Who are the vault providers and application vault keepers, and what are the timelocked windows? The proposal does not disclose them. An actor set whose members are unknown is a trust assumption you cannot evaluate.
  • What is the universal challenger threshold, and can a depositor realistically operate as their own challenger? The proposal says a depositor can act as a challenger by operating their own vault provider. Verify what that actually requires in practice, because “you can be your own validator” is a sentence that has been used to describe a lot of systems nobody self-validates.
  • What exactly can the 2-of-3 multisig change, and how fast? The sell discount on liquidations and the arbitrageur margin are direct value transfers from borrowers. A timelock is a delay. Know the delay length.
  • What happens to a position if Babylon governance exercises a pause or freeze? Your BTC is locked in a Taproot UTXO whose redemption depends on actors. If those actors are paused, what is your path out, and how long does it take?
  • How does the WBTC liquidation payment behave if WBTC itself depegs or loses liquidity? Your collateral is bitcoin; your liquidation settles in a wrapped token. That is a new form of basis risk that does not exist in a single-chain lending market.

The summary I would offer, borrowing a framing that has outlasted most crypto infrastructure: this proposal does not eliminate trust. It relocates trust from a custodian you can sue to a set of actors you cannot name, governed by three committees, with a liquidation path that pays you in a different asset than the one you posted. Every one of those may be a reasonable trade. But a trade is not an absence, and the word “trustless” is doing a lot of unearned work in the announcement.

Chuang Tzu observed that a summer insect cannot be shown ice, because it has no experience of frost, and a well-frog cannot discuss the sea, because it is confined to its own hollow. The custodial model was our frost. We have been arguing about whether the new design is warmer, without asking whether it is simply a different season with a different kind of cold.

(The End)

Blockchain

Your Seed Phrase Never Left the Screen. That's How It Was Stolen.

2026-10-15 2:55:03

Blockchain

Six Confirmations Is a Price Tag, Not a Receipt: What a 51% Attack Actually Buys

2026-10-17 13:25:21

❯
Profile
Cart
Coupons
Search