# Bitcoin-IPC: Scaling Bitcoin with a Network of Proof-of-Stake Subnets

Crypto.world

> Bitcoin holds over $1.7T in value but its L1 processes only a few transactions per second, and today's L2 ecosystem is fragmented: value in one L2 cannot move to another without an external bridge or…

**arXiv**: 2512.23439

**Published**: 2025-12-29

**Authors**: Orestis Alpos, Jakov Mitrovski, Themis Papameletiou, Nikola Ristić, Dionysis Zindros, Marko Vukolić

## AI summary

## Overview

**Research area:** Bitcoin scaling and interoperability — Layer-2 (L2) design, proof-of-stake (PoS) subnet architectures, cross-chain settlement, and Bitcoin Script engineering.

**Technical level:** Advanced. The paper assumes familiarity with UTXOs, Taproot/script trees, sighash flags, BFT consensus, PoS slashing, and long-range attacks.

**Scope:** A protocol design, formalization, and benchmark study of Bitcoin-IPC, a network of permissionless PoS L2 subnets that use Bitcoin L1 for staking, settlement, and inter-subnet transfers without external bridges or a Bitcoin consensus change.

## What This Paper Is About

Bitcoin holds over $1.7T in value but its L1 processes only a few transactions per second, and today's L2 ecosystem is fragmented: value in one L2 cannot move to another without an external bridge or swap. The paper proposes Bitcoin-IPC, a network of programmable PoS subnets whose validators stake BTC on Bitcoin L1, and whose cross-subnet transfers are batched into Bitcoin transaction witness data and settled by Bitcoin itself. The goal is to horizontally scale Bitcoin's monetary throughput while keeping each subnet anchored to, and protected by, Bitcoin L1.

## Key Contributions

1. **A permissionless, interconnected subnet architecture for Bitcoin.** Any BTC holder can create a subnet; validators stake BTC as collateral and run a BFT consensus weighted by stake. Inter-subnet transfers involve only the source and destination subnet validators plus Bitcoin L1, with no external bridges and no per-path or per-subnet liquidity lock-up.
2. **A Bitcoin-native encoding of the full subnet lifecycle.** The `create`, `join`, `leave`, `stake`, `kill`, and `getSubnet` operations, plus `deposit`, `withdraw`, and `transfer`, are implemented as ordinary Bitcoin transactions using two data-attachment mechanisms: `Op_Return` (limited to 80 bytes per transaction) for small payloads, and commit-reveal inside the witness (up to Bitcoin's maximum standard transaction size of 100K virtual bytes, at the 4x discounted fee rate) for larger ones. No Bitcoin consensus update is required, and standard Bitcoin nodes can simply ignore these transactions.
3. **Bitcoin-rooted security properties with formal treatment.** BTC collateral provides Sybil protection; equivocating validators are slashed using Babylon's EOTS primitive adapted to a dynamic, multi-subnet setting with no central L2; honest validators get unilateral exit; periodic anchoring bounds PoS long-range attacks; and UTXO conservation gives a firewall property so a compromised subnet cannot extract more BTC than was deposited into it. These are formalized in Appendix B.
4. **A batching design that amortizes cross-subnet transfers.** Because cross-subnet settlements are embedded and batched in Bitcoin's witness mechanism, they approach 6 vB per transfer — 23x less than a native L1 payment — which, when Bitcoin blocks are filled with such batches, corresponds to about 273 transfers per second.

## Main Findings

- **Baseline L1 throughput is low, and only modestly improvable.** The commonly cited 7 tps corresponds to an average transaction size of about 238 vB. Using the 141 vB size of a baseline "send + change" transaction produced by modern wallets, 1M vB per block, and one block every 10 minutes, the paper computes a baseline throughput of 11.8 tps — still insufficient for global use.
- **Cross-subnet transfers drop to roughly 6 vB.** Batching many transfers into witness data approaches 6 vB per transfer in the large-batch limit, versus 141 vB for a typical native payment, a 23x reduction in on-chain size and cost.
- **About 273 transfers per second is the implied L1 monetary throughput.** Filling Bitcoin blocks with such batches yields roughly 273 transfers per second, 23x the baseline, without modifying Bitcoin.
- **Intra-subnet transfers scale independently of L1.** They remain on separate paths and scale horizontally, so L1 throughput is not the ceiling for activity inside a subnet.
- **Security degrades gracefully with the Byzantine stake fraction F_e.** If F_e ≤ 1/3 for every epoch, the subnet is correct (1/3-bounded). If 1/3 < F_e ≤ 2/3 for some epoch, safety violations may occur but imply attributable, slashed equivocation; liveness violations may occur (for example stalling `join` of new validators and checkpoints), but funds are never stolen since spending requires more than 2/3 of the stake. If F_e > 2/3, the subnet is compromised with no guarantees to its users. The protocol makes no assumption on the number of Byzantine validators across subnets as a whole.
- **Collected collateral does not count toward security.** Subnet security does not depend on validators that have withdrawn collateral, which is why long-range attacks are addressed through a Bitcoin-anchored, tamper-evident history rather than through slashing.
- **Deposits and withdrawals are bridge-free.** Deposits lock BTC under the current configuration's multisig and mint an equal amount of wBTC on the subnet; withdrawals burn wBTC and release BTC on Bitcoin — using only the communicating subnets' validators and Bitcoin.
- **The `create`/`join` reveal transactions are front-runnable but harmless.** Once broadcast, the data script contents are public and anyone can spend the commitment output, but those transactions transfer no value and any replacement must disclose the same committed parameters.
- **Detailed benchmark results are not reported in the provided content.** Section 5 is described as containing the benchmarks, but the excerpt above provides only the 6 vB, 23x, 11.8 tps, and 273 transfers-per-second figures.

## Methodology in Plain English

The authors take an existing horizontal-scaling paradigm — PoS sidechains interconnected through a root chain — and work out how to instantiate it on Bitcoin, which has no general-purpose computation. Their approach has four parts.

First, they define a small grammar of Bitcoin locking scripts (signature checks, weighted multisig satisfied by more than 2/3 of total weight, absolute and relative timelocks, and boolean combinations) and model Bitcoin as a synchronous network where transactions are considered final after six confirming blocks. On top of that they specify the protocol interface and the security properties they want.

Second, they encode every protocol operation as concrete Bitcoin transactions. Small metadata goes into an 80-byte `Op_Return` output; larger payloads use a commit-reveal pair, where one transaction commits to a data script (pushing data in chunks of at most 520 bytes, then dropping it and pushing true) and a second transaction reveals it in the witness, taking advantage of the witness discount. Collateral is locked in a three-leaf script tree: a unilateral-exit leaf spendable by the owner alone after a timeout, a slash leaf, and a cooperative-exit leaf requiring both the owner and more than 2/3 of the configuration's stake. A separate unbonding script tree keeps a cooperatively exiting validator exposed to slashing for a relative timelock of three epochs (3K blocks).

Third, they design slashing around equivocation. Validators sign subnet blocks with a Schnorr key and per-height nonces committed in advance via `join` and `stake` transactions; signing two conflicting blocks at the same height reuses a nonce, which reveals the secret key. The active configuration pre-signs a slashing transaction when collateral is created, publishing it through checkpoints, so anyone who extracts the leaked key can complete the signature and burn the collateral.

Fourth, they argue security informally in the body and formally in the appendices, and evaluate the on-chain footprint of batched transfers to derive the vB-per-transfer and throughput figures.

## Why This Matters

**Impact on research.** The paper shows that Bitcoin's existing scripting capabilities — script trees, timelocks, weighted multisig, witness data — are sufficient to build a staked, slashing-secured, interoperable L2 network without a soft fork or hard fork. It also extends Babylon-style non-custodial BTC staking from a single trusted proxy chain to a permissionless, dynamic, multi-subnet setting, and gives a concrete Bitcoin instantiation of the firewall property previously studied for contract-capable L1s.

**Real-world applications:**

- Payments: batched cross-subnet settlement at roughly 6 vB per transfer, with intra-subnet transfers off the L1 path entirely.
- Programmable Bitcoin finance: each subnet runs its own programmable engine (for example, the EVM), enabling smart contracts backed by BTC collateral.
- Moving value between L2s: users can move wBTC across subnets through Bitcoin rather than trusting a third-party bridge or accepting per-path liquidity lock-up.
- Permissioned consortium or application-specific chains that still settle to Bitcoin and can be terminated through the `kill` operation, with a defined wind-down window for users to withdraw.

**Industry relevance.** The design speaks directly to the fragmentation problem in the current Bitcoin L2 landscape — hundreds of proposals with far fewer live chains, most isolated from one another. It offers a path to interoperability that requires no new Bitcoin opcodes, uses BTC rather than a separate token for voting weight and slashing, and is published as open source.

## Future Directions

- **Formal verification and refined proofs.** The appendices contain the complete formalization and proofs; extending these to cover the full dynamic, multi-subnet setting — membership churn, delayed arrivals, and cross-subnet composition — is a natural next step.
- **Incentive design.** Staking incentives are treated as optional in Section 4; how subnets should reward validators, and how rewards interact with the decision to join or leave, remains an open design question.
- **Parameter selection and operational experience.** Parameters such as the minimum collateral, checkpoint interval, epoch length, and the `kill` proposal and wind-down windows are set per subnet at creation; guidance on choosing them, and measurements from live deployments, are not reported.
- **Bootstrap and adoption.** The initial configuration `C_init` is explicitly a bootstrap mechanism needed before slashing and collateral scripts have a concrete validator set to name; how subnets credibly seed that set and transition to permissionless membership is left to future work.

## Target Audience

Protocol researchers and engineers working on Bitcoin scaling, L2 interoperability, and proof-of-stake security; implementers of Bitcoin Script and Taproot-based protocols; and technically sophisticated readers who want to understand how staking, slashing, and cross-chain settlement can be built on Bitcoin without changing its consensus rules. Readers without a background in UTXO scripting and BFT consensus will find the paper demanding.

## Authors’ abstract

Despite holding over…

[The full abstract on arXiv](https://arxiv.org/abs/2512.23439)

[Read the original paper](https://arxiv.org/abs/2512.23439)

---

Canonical: https://crypto.world/paper/bitcoin-ipc-scaling-bitcoin-with-a-network-of-proof-of-stake-subnets-2512.23439
