#baby $BABY @BabylonLabs_io spent the morning trying to understand what makes @BabylonLabs_io’s trustless vault design expensive, and somehow ended up stuck on a hard-drive number.
The whitepaper’s BitVM3 benchmark says generating one garbled circuit takes around 20 minutes on a single CPU core.
Storage per counterparty: 43 GB.
That is not Bitcoin transaction data sitting onchain. It is off-chain challenge material each side may need if someone later submits an invalid proof. Large borrowers can generate and store it themselves. The paper expects smaller borrowers will probably hand that job to professional operators.
That’s the part that made me pause.
The operator still cannot simply take the BTC. The borrower co-signs every transaction capable of spending it, so outsourcing the circuit work does not recreate a custodian.
But it does create a different kind of dependence.
A smaller user may technically control the Bitcoin while relying on someone else to generate the machinery, store it correctly and keep it available in case a dispute actually happens.
Made coffee, came back and realised I had been treating self-custody and self-operation like the same thing.
They aren’t.
Maybe this is completely normal. Most people already use hosted infrastructure instead of running every node, prover or indexer themselves. Professional operators may be what turns a technically trustless design into something ordinary users can actually access.
Still, the distinction matters.
Babylon may remove the party that can steal your BTC without removing the party you depend on to keep the protection system practical.
So when a borrower outsources 43 GB of challenge material, is the vault still fully trustless for that user — or merely non-custodial with an operational service wrapped around it?
@BabylonLabs_io
$BABY #Babylon