Zest's 'BTC Never Leaves Bitcoin' Loan Went Live on 23 Sept but Is Capped at 0.001 BTC (~$85) Per Wallet: Phase 1 Is a Two-Key Taproot Vault Policed by Watchtowers and a Named Guardian Council, Not BitVM; Zest's Real Book Is $75.8M on Stacks Sitting Behind sBTC's 15-Signer, 70% Threshold Peg
What the headline says versus what shipped. The parent card describes Zest as 'developing' a protocol that 'would allow' BTC to collateralise loans without leaving Bitcoin. That is out of date by about a day: on 23 September 2026 Zest published 'Bitcoin Collateral Vaults mainnet demo is live' (zestprotocol.com blog, 23 Sept). Native BTC goes into a Taproot vault on Bitcoin L1 and USDC is borrowed on Ethereum, not on Bitcoin and not on Stacks. The catch is in the same post: the per-wallet collateral cap is 0.001 BTC, roughly $85 at the $85,450 CoinGecko print on 23 Sept, and it stays capped until external audits complete. This is a mainnet proof-of-mechanism, not a lending market. The correct category is defi, not price; nothing here moves spot.
The mechanism is not BitVM yet. Zest's own docs ('How Zest Protocol brings Bitcoin Collateral Vaults to mainnet', docs.zestprotocol.com) describe Phase 1 as a two-key vault: the depositor's key plus a protocol key that is split across operators by threshold signing. Every spend path (reclaim to depositor, liquidation to a registered liquidator, an emergency path to a hardcoded depositor failsafe) is co-signed by the depositor at deposit time, so 'nobody can add a destination afterwards'. Three off-chain roles then decide which pre-signed path becomes valid: 'watchtowers' (a supermajority of distributed operators who publish observed Ethereum loan state), a 'Guardian Council' of named institutional parties on a conventional multisig whose keys can only route funds back to the depositor, and the liquidator. A repaid borrower can sweep unconditionally after a one-day Bitcoin timelock if the council is unresponsive. That is a real improvement over a custodial peg: no coalition of operators can steal the coins, because every exit is missing the depositor's signature. But it is still a federation deciding whether a liquidation fires. Phase 2, per the same page, moves that authority to 'a cryptographic proof of the lending position's state, verified on Bitcoin via BitVM's optimistic challenge protocol', with the challenger role opened to everyone. Phase 2 has no date.
What BitVM will and will not fix. BitVM2's own design page (bitvm.org/bitvm2) states the bridge assumption plainly: among the operators 'at least one of them has to act honest', 'otherwise the funds become unspendable eventually', while 'anyone can challenge an invalid assertion without having to be part of the initial group of n'. So the end state is 1-of-n liveness rather than trustlessness, plus Ethereum consensus for the loan state and an oracle for the BTC price that neither Zest page specifies. Readers should treat 'enforced by Bitcoin's own rules' as a Phase 2 aspiration.
Base rates: where Zest's money actually is. DefiLlama's protocol feed (api.llama.fi/protocol/zest, fetched 23 Sept 12:12 UTC) puts Zest at $75.82M TVL, all on Stacks, with $15.00M borrowed; the all-time high was $91.76M on 3 Oct 2025. At $85,450 that TVL is about 887 BTC-equivalent, consistent with the '800+ BTC deposited' and '1,500+ liquidations with zero bad debt' Zest claimed in its 6 May 2026 Chainwire release announcing the vault prototype. That book runs on sBTC, and sBTC is exactly the model the new vault is meant to escape: per docs.stacks.co, all backing BTC sits in 'a single Bitcoin UTXO (the peg wallet)' controlled by '15 community-chosen signers' at a '70% consensus threshold', with peg-outs needing Stacks finality plus six Bitcoin confirmations. Zest is therefore 84% of the entire Stacks chain ($89.75M, DefiLlama chains endpoint, 23 Sept), and the vault, at 0.001 BTC per wallet, is a rounding error against it.
Scale of the opportunity the card gestures at. DefiLlama counts $4.57B of TVL on the Bitcoin chain itself on 23 Sept, but $3.46B of that (76%) is Babylon restaking, not lending. Rootstock is $89.7M, BOB $10.4M, Merlin $10.0M, Core $5.4M, Bitlayer $0.5M. Ethereum is $54.26B and WBTC alone is a $9.99B bridge. So 'Bitcoin DeFi' outside Babylon is roughly $1.1B on L1 plus about $0.2B across the sidechains and L2s, versus $10B of BTC that holders were willing to hand to a custodian to use on Ethereum. That gap is the demand signal for a design where the coin stays on Bitcoin; the same gap is the evidence that custody trust has not been the binding constraint for the marginal user, yield and liquidity have.
What would change the read. Bullish for the thesis: the cap lifting after audits with disclosed liquidator and oracle parameters; a dated Phase 2 with a live BitVM challenge on mainnet; vault TVL above 100 BTC. Bearish: a Guardian Council intervention that reverses a liquidation (it would prove the federation, not Bitcoin, is the arbiter), or the Stacks market shrinking further from its Oct 2025 high while the L1 product stays a demo. Net: this is a credible engineering milestone with a sub-$100 ceiling, and the parent card's enthusiasm should be read against $15M of actual Zest borrowing, all of it on a 15-signer peg.
Sources (6)
AI Research
Key Takeaway
Zest's native-BTC collateral vault is live on mainnet as of 23 Sept 2026 but capped at 0.001 BTC per wallet and secured by watchtowers plus a named Guardian Council rather than BitVM; Zest's actual $75.8M book and $15.0M of borrowing still sit on Stacks behind sBTC's 15-signer, 70%-threshold peg.
What the headline says versus what shipped. The parent card describes Zest as 'developing' a protocol that 'would allow' BTC to collateralise loans without leaving Bitcoin. That is out of date by about a day: on 23 September 2026 Zest published 'Bitcoin Collateral Vaults mainnet demo is live' (zestprotocol.com blog, 23 Sept). Native BTC goes into a Taproot vault on Bitcoin L1 and USDC is borrowed on Ethereum, not on Bitcoin and not on Stacks. The catch is in the same post: the per-wallet collateral cap is 0.001 BTC, roughly $85 at the $85,450 CoinGecko print on 23 Sept, and it stays capped until external audits complete. This is a mainnet proof-of-mechanism, not a lending market. The correct category is defi, not price; nothing here moves spot.
The mechanism is not BitVM yet. Zest's own docs ('How Zest Protocol brings Bitcoin Collateral Vaults to mainnet', docs.zestprotocol.com) describe Phase 1 as a two-key vault: the depositor's key plus a protocol key that is split across operators by threshold signing. Every spend path (reclaim to depositor, liquidation to a registered liquidator, an emergency path to a hardcoded depositor failsafe) is co-signed by the depositor at deposit time, so 'nobody can add a destination afterwards'. Three off-chain roles then decide which pre-signed path becomes valid: 'watchtowers' (a supermajority of distributed operators who publish observed Ethereum loan state), a 'Guardian Council' of named institutional parties on a conventional multisig whose keys can only route funds back to the depositor, and the liquidator. A repaid borrower can sweep unconditionally after a one-day Bitcoin timelock if the council is unresponsive. That is a real improvement over a custodial peg: no coalition of operators can steal the coins, because every exit is missing the depositor's signature. But it is still a federation deciding whether a liquidation fires. Phase 2, per the same page, moves that authority to 'a cryptographic proof of the lending position's state, verified on Bitcoin via BitVM's optimistic challenge protocol', with the challenger role opened to everyone. Phase 2 has no date.
What BitVM will and will not fix. BitVM2's own design page (bitvm.org/bitvm2) states the bridge assumption plainly: among the operators 'at least one of them has to act honest', 'otherwise the funds become unspendable eventually', while 'anyone can challenge an invalid assertion without having to be part of the initial group of n'. So the end state is 1-of-n liveness rather than trustlessness, plus Ethereum consensus for the loan state and an oracle for the BTC price that neither Zest page specifies. Readers should treat 'enforced by Bitcoin's own rules' as a Phase 2 aspiration.
Base rates: where Zest's money actually is. DefiLlama's protocol feed (api.llama.fi/protocol/zest, fetched 23 Sept 12:12 UTC) puts Zest at $75.82M TVL, all on Stacks, with $15.00M borrowed; the all-time high was $91.76M on 3 Oct 2025. At $85,450 that TVL is about 887 BTC-equivalent, consistent with the '800+ BTC deposited' and '1,500+ liquidations with zero bad debt' Zest claimed in its 6 May 2026 Chainwire release announcing the vault prototype. That book runs on sBTC, and sBTC is exactly the model the new vault is meant to escape: per docs.stacks.co, all backing BTC sits in 'a single Bitcoin UTXO (the peg wallet)' controlled by '15 community-chosen signers' at a '70% consensus threshold', with peg-outs needing Stacks finality plus six Bitcoin confirmations. Zest is therefore 84% of the entire Stacks chain ($89.75M, DefiLlama chains endpoint, 23 Sept), and the vault, at 0.001 BTC per wallet, is a rounding error against it.
Scale of the opportunity the card gestures at. DefiLlama counts $4.57B of TVL on the Bitcoin chain itself on 23 Sept, but $3.46B of that (76%) is Babylon restaking, not lending. Rootstock is $89.7M, BOB $10.4M, Merlin $10.0M, Core $5.4M, Bitlayer $0.5M. Ethereum is $54.26B and WBTC alone is a $9.99B bridge. So 'Bitcoin DeFi' outside Babylon is roughly $1.1B on L1 plus about $0.2B across the sidechains and L2s, versus $10B of BTC that holders were willing to hand to a custodian to use on Ethereum. That gap is the demand signal for a design where the coin stays on Bitcoin; the same gap is the evidence that custody trust has not been the binding constraint for the marginal user, yield and liquidity have.
What would change the read. Bullish for the thesis: the cap lifting after audits with disclosed liquidator and oracle parameters; a dated Phase 2 with a live BitVM challenge on mainnet; vault TVL above 100 BTC. Bearish: a Guardian Council intervention that reverses a liquidation (it would prove the federation, not Bitcoin, is the arbiter), or the Stacks market shrinking further from its Oct 2025 high while the L1 product stays a demo. Net: this is a credible engineering milestone with a sub-$100 ceiling, and the parent card's enthusiasm should be read against $15M of actual Zest borrowing, all of it on a 15-signer peg.