State of Layer 2
Ark reaches mainnet without covenants, BitVM bridges become the federations' stated successor, and chaumian ecash moves on-chain.
Takeaway
A spectrum from most-trustless to most-federated: Ark and covenants, BitVM bridges, ecash, statechains, and one sidechain winding down.
Ark Reaches Mainnet
A shared-UTXO design that ships today without any new consensus rules.
THE LAUNCH
Second (Steven Roose) put its Ark client bark on Bitcoin mainnet on 9 June, which Second describes as the first Ark client on mainnet rather than signet.
WHAT ARK IS
Many users share on-chain UTXOs through trees of pre-signed off-chain transactions (VTXOs). There are no Lightning channels and no pre-allocated liquidity, and a user can still exit unilaterally to the main chain.
NO COVENANTS NEEDED
bark runs without CTV or CSFS today, using all-of-all pre-signed "pseudo-covenants". Covenants would make it cheaper and less interactive, but are not required to ship.
TWO ARKS, DON'T CONFLATE
Second builds bark / barkd / aspd (the mainnet launch). Ark Labs builds Arkade / arkd, in public beta since October 2025 with point releases this month. Different teams, same idea.
What Ark Is For
The clearest use case is cheap, self-custodial payments without channel liquidity.
AFRICA SETTLEMENT PILOT
At the bitcoin++ Nairobi hackathon, Second, Tando, Mavapay and JunubBTC ran a pilot where Tando swapped its phoenixd Lightning setup for a Bark client, cutting liquidity cost an estimated 50-80% while keeping self-custodial, exitable VTXOs.
ECASH TO MOBILE MONEY
The pilot melts blinded ecash into a mobile-money gateway over Ark at 0% network fee, with a Bark cosignature proof-of-reserves. It runs on testnet (Kenya M-PESA to South Sudan MTN MoMo), separate from the mainnet launch.
LIGHTNING BACKEND
Alby built a Bark backend for its hub, so NWC apps can make Lightning payments backed by Ark VTXOs with no Lightning node or channels of their own.
SECURITY RELEASE TRAIN
Six bark releases (0.2.2 to 0.3.0) focused on the unilateral-exit game: 0.2.5 was a security release (nonce-reuse and HTLC-expiry fixes), and 0.3.0 stops committing VTXOs until broadcast and auto-recovers force-exited ones.
Second's launch post omitted round interval, VTXO expiry, exit window and amount caps; confirm parameters before relying on them.
Covenants Without a Soft Fork
Two June threads showed how much is already possible on today's rules.
CSFS + CAT DEMOS
AaronZhang posted three live Inquisition-signet transactions showing OP_CHECKSIGFROMSTACK + OP_CAT can reproduce most dedicated covenant behaviour: CTV-style output binding, input binding (constraining
sha_prevouts), and eltoo-style state replacement. He concedes it shows capability, not efficiency.TRANSACTION EXPIRY VIA BIT 21
Josh proposed giving nSequence bit 21 a meaning: such a transaction expires if its input confirms too late. A 100-block maturity rule makes it immune to the free-relay attacks that sank OP_EXPIRE, and it defeats Lightning replacement-cycling with "no new opcode, just bit 21".
BitVM Bridges
Federated pegs are naming BitVM-style bridges as their intended successor.
UNION ON ROOTSTOCK
Fairgate Labs and RootstockLabs put the Union bridge on Rootstock testnet (on BitVMX): lock BTC on L1, mint on Rootstock, redeem via on-chain fraud proofs, secured by a 1-of-n honest functionary set. This is the path Rootstock intends to use to replace its 5-of-9 federated PowPeg.
LIQUID NAMES A SUCCESSOR
Blockstream's roadmap names a BitVM-style 1-of-n bridge as the eventual successor to Liquid's 15-functionary multisig peg. It is research-stage, but the oldest production sidechain federation naming BitVM is a directional signal.
WHERE THE DEBATE IS
The open question has shifted from "does it work" to the griefing and DoS economics of operator bonds and challenge games.
DON'T MISREAD CITREA
Citrea widening a bridge multisig from 2-of-2 to 3-of-4 is a custody-multisig change for wrapped assets, not a milestone for its BitVM2 Bitcoin bridge. Keep the two separate.
Ecash As a Bitcoin Layer
Chaumian ecash spec work is making mints usable as on-chain on-ramps.
NUT-30: ON-CHAIN MINT/MELT
Cashu's NUT-30 lets you mint and melt ecash directly against on-chain Bitcoin with no Lightning hop, using a reusable deposit address, so a mint can act as an on and off ramp. Shipped in CDK v0.17.0.
NUT-12: SAFER NONCES
NUT-12 derives the DLEQ proof nonce deterministically via HMAC instead of trusting the RNG, closing a nonce-reuse to mint-key-recovery footgun, the same failure mode as the 2013 repeated-nonce ECDSA thefts.
PAYJOIN TO CASHU
A draft (CDK PR #2049) combines NUT-30 on-chain deposits with BIP-77 async Payjoin: the receiver contributes a UTXO to break the common-input heuristic.
FEDIMINT V2
Fedimint's v2 mint and wallet modules now compile under WASM for browser and mobile clients, with a gateway latency win. Still experimental on master.
Statechains: Spark and Mercury
Fast shipping does not change the underlying trust model; read it directly.
SPARK SHIPPED A LOT
Lightspark's Spark added a two-phase-commit consensus engine, mutual operator authentication, exit and refund hardening, and five SDK releases this month.
THE TRUST MODEL
Spark is a two-operator federation (Lightspark and Flashnet) whose safety rests on honest key deletion at transfer time that cannot be observed or proven. Spark's own research brands this "perfect forward security"; read it as a semi-trusted federation with a liveness-dependent exit and judge the framing yourself.
MERCURY LAYER
Mercury Layer is the blinded-operator contrast, where the operator cannot see the transfer tree. As of late June its public repository was not reachable without authentication and had no recent releases; treat its status as unconfirmed.
A TEACHING PAIR
Blinded operator (Mercury) versus sees-the-tree operator (Spark) is a clean way to show what "statechain" actually leaves you trusting.
A Sidechain Winds Down
An orderly shutdown still ends in a federation sweep at a deadline.
BOTANIX CLOSES SPIDERCHAIN
Botanix is winding down its EVM Spiderchain. Users must peg out before 9 July; after that the rotating BTC-bonded federation sweeps remaining BTC, and other tokens are unrecoverable.
WHY IT MATTERS
It ran about a year (~25M transactions, ~200k wallets) and shut down over economics, not an exploit: fee revenue never covered costs and there was no token to subsidise it. A rare clean post-mortem of a federated-peg sidechain.
Discussion
Which of these is actually trust-minimised?
QUESTION
For each design, who can take or freeze your coins, and what does a unilateral exit really require?
QUESTION
Ark ships without covenants today. What would CTV or CSFS actually buy it, and is that worth a soft fork?
QUESTION
If federations are naming BitVM bridges as their successor, how far off is a 1-of-n bridge you would trust with real size?
OPEN ISSUE
Several projects market honest-operator or custodial-multisig designs as trustless L2s; read the trust model before repeating the label.