Date: September 2025 · Authors: Paragon Labs Core Contributors

Abstract

Paragon Chain is a high-performance Layer-1 blockchain that integrates an Aptos-derived MoveVM architecture with native modules for AI compute and compliant real-world asset (RWA) tokenization. It supports modular roll-ups settling to a unified state, aiming for a secure, horizontally scalable, developer-first network with sustainable economics.

• AI Compute Module (AICM): On-chain escrow, coordination, and proof market for AI inference and training jobs.
• RWA Registry: Move-based framework for compliant tokenization with verifiable off-chain attestations.
• Modular L2 Manager: Registration, bridging, and fee-sharing for MoveVM roll-ups and zkVM side-chains.
• Dynamic Gas & Revenue-Sharing: EIP-1559-style base fees with programmable splits to validators, dApps, and DAO public goods.
• XPGN Tokenomics: 550M cap, validator staking incentives, and community emissions designed for long-term sustainability.

1. Introduction

Web3 has matured into a general-purpose economic operating system, yet developers grapple with high fees, fragmented liquidity, and constrained on-chain compute. Paragon Chain tackles these challenges by extending an Aptos-inspired parallel-execution foundation with integrated AI and RWA modules, alongside a design optimized for roll-ups. To bootstrap adoption, we launch pragmatically with Paragon DEX on BNB Chain, fostering real users and liquidity before transitioning to the L1 as community needs evolve.

2. Vision

• Developers deploy AI-powered, RWA-backed DeFi applications with predictable costs and formally verified safety.
• Users transfer assets seamlessly across shards, L2s, and partner chains with fast, low-fee finality.
• Validators earn sustainable rewards from gas fees, MEV sharing, and modular fee markets, avoiding hardware-intensive centralization.
• The DAO governs upgrades and tokenomics transparently, minimizing disruptive hard forks.

3. Architecture Overview

3.1 Base Layer (Aptos-Derived MoveVM)

Paragon inherits Aptos' pipelined transaction lifecycle: dissemination, BFT metadata ordering (HotStuff variant), Block-STM parallel execution, batched writes, and ledger/state certification. This enables high throughput, with Aptos demonstrating up to 160,000 TPS in controlled environments and real-world peaks around 13,367 TPS without failures or gas spikes.

3.2 Key Extensions

4. Smart-Contract Model: Move++

Move++ extends the Move language with features for concurrency and modularity:

• Actor Interfaces: Define capabilities for roles like AI workers, RWA attestors, and bridge relayers. Example pseudocode for an AI worker registration:

move module AICM::AIWorker {
  struct WorkerCapability has key { id: u64, stake: Coin<XPGN> }

  public fun register_worker(account: &signer, stake_amount: u64): WorkerCapability acquires Coin {
    let stake = coin::withdraw<XPGN>(account, stake_amount);
    move_to(account, WorkerCapability { id: account::new_id(), stake });
    WorkerCapability { id, stake }
  }

  public fun submit_proof(cap: &mut WorkerCapability, job_id: u64, zk_proof: vector<u8>) {
    // Verify zk-SNARK and settle escrow
    assert!(verify_zk_snark(zk_proof, job_id), E_INVALID_PROOF);
    // Release stake and rewards
  }
}

• Delta Resources: Support additive writes (Δ) to minimize storage hotspots in high-concurrency scenarios.
• Type-Safe External Calls: Enable cross-shard invocations with canonical schemas for atomic composability.

5. Data Model & Storage

Paragon maintains a global address space with Sharded Tables: Merkle-aggregated key-value stores across shards hashing to a unified state root. Storage rent, paid in XPGN, is auto-rebated upon pruning. For cross-shard atomicity, we employ a two-phase commit: prepare deltas with locks across shards, then commit or rollback via compensating transactions. Failures trigger Merkle-proof-based rollbacks, ensuring consistency without halting the chain.

6. Security & UX

• Multi-signature/MPC account recovery in the standard library.
• Pre-execution simulation API from full nodes for wallet previews of transaction effects.
• Adaptive risk alerts for AI jobs and unverified RWAs.

7. Pipeline Enhancements

• Speculative read/write hints project a ~40% reduction in Block-STM re-executions on mixed workloads.
• libp2p dissemination with erasure-coded batches reduces bandwidth compared to baseline Aptos.
• Optional accelerator hooks for hashing/signatures, avoiding consensus bias.

Benchmarking roadmap: Pre-testnet simulations on EC2 instances will validate these on mixed DeFi/AI workloads, publishing open-source results.

8. AI Compute Module (AICM)

• Lifecycle: Create → Stake → Work → Submit Proof → Verify → Settle.
• Proofs: zk-SNARKs for deterministic inference; committee signatures for stochastic training.
• Use cases: Personal agents, real-time market-makers, research compute.
• Differentiation: Unlike zkML-focused projects (e.g., Modulus Labs), Paragon embeds AICM natively in the L1 for seamless RWA-AI composability without cross-chain overheads; hybrid proofs balance deterministic/stochastic tasks.

9. RWA Registry & Compliance

Assets: equity, debt, real estate, commodities, carbon credits.

Compliance: whitelists, jurisdiction tags, rule-based transfers; attestor quorums + CCIP oracles.

Differentiation: While Centrifuge focuses on invoices/receivables and EVM integrations, Paragon's Move-based registry emphasizes formal verification and AI synergies (e.g., AI-driven valuations), remaining asset-agnostic.

10. Modular L2 & zk Support

• Registration transactions mint an L2Handle resource with escrowed XPGN bonds.
• Canonical light-client contracts target <2-block finality.
• Up to 50% of L2 gas rebates to L1 validators for incentive alignment.

11. XPGN Token Economics

The DAO may extend supply beyond the cap with >66% vote and 6-month timelock. Emissions prioritize community incentives, validator security, and public goods. For resilience, treasury diversification (e.g., 20% in stables) and stress tests simulate 90% market drops.

12. Risks and Mitigations

• AI Proof Centralization: Committee collusion mitigated by VRF-based random selection and slashing; transition to broader zk as costs fall.
• zk-SNARK Reliance: High proving costs offset via off-chain generation + on-chain verification.
• Sharding Fragmentation: Two-phase commits + audits; Merkle-proof-based rollbacks for consistency.
• Regulatory Risk: Evolving RWA laws; DAO-driven parameter updates and disclosures.
• Adoption Risk: Bootstrap via BNB DEX, hackathons, and partnerships.

13. Performance & Sharding Roadmap

Targets reference Aptos benchmarks (160k TPS tests) and project scaling via sharding to potentially 500k+ TPS.

14. Phased Launch & Ecosystem Growth

• Phase 0 — DEX & Vaults (BNB): Launch AI-powered liquidity pools and RWA vaults; distribute XPGN via airdrops/hackathons.
• Phase 1 — Public Testnet: Validator incentives, tooling, explorer, bridges.
• Phase 2 — Mainnet L1: MoveVM core, EVM↔Move bridges, AICM/RWA modules, DAO activation.

Ecosystem: Target partnerships (e.g., grants, bridges); host AMAs on X/Discord for feedback. Initial dApps focus on DeFi-AI hybrids.

15. Conclusion

Paragon combines Move-grade performance with native AI markets and RWA tools. Through modularity, verification, and sustainable economics, it powers the next era of on-chain finance, data, and intelligence.

References

• Aptos Labs — The Aptos Blockchain: Safe, Scalable & Upgradeable Web3 Infrastructure (2022).
• Blackshear et al. — Move: A Language with Programmable Resources (2019).
• Additional sources cited inline.