Arcium is a decentralized network designed to support encrypted computations, focusing on providing privacy for on-chain applications and offering a verifiable, trustless framework for developers.^[1][2]

Overview

Arcium, founded in 2022 by Yannik Schrade, Julian Deschler, Nicolas Schapeler, and Lukas Steiner, aims to provide a decentralized confidential computing network. Initially called Elusiv, it operates a zero-knowledge privacy protocol on the Solana blockchain and has secured $9 million in funding across two rounds, including a strategic round led by Greenfield Capital.

In May 2024, Arcium launched its incentivized private testnet as part of its roadmap. The network is designed for secure data collaboration in sectors like blockchain, healthcare, and AI, using a distributed node architecture to perform Multi-Party Computation (MPC) tasks while maintaining data encryption.

Arcium's framework on the Solana blockchain handles tasks such as computation scheduling and compensation. It aims to offer flexible MPC configurations, allowing users to tailor security and execution protocols for applications like AI model training and confidential data analysis.^{[1][2][3][4][5]}

Architecture

Arcium’s architecture is designed to support distributed confidential computing. It breaks tasks into computation definitions, which are executed in Multi-Party Computation (MPC) environments (MXEs) on clusters of Arx nodes, functioning like hardware for secure processing.

The network is coordinated by programs running on the Solana blockchain, aiming to manage tasks through an on-chain mempool system. Compensation for computational services is handled on-chain, facilitating payments from customers to Arx operators and third-party delegators.^[6]

Products

Multi-Party Execution Environments (MXEs)

Multiparty Execution Environments (MXEs) aim to facilitate secure computations on encrypted data, enabling collaboration among multiple parties without revealing their inputs or outputs. Developers can customize the encryption schema based on the sensitivity of the data, generating a shared key for participating nodes.

MXEs serve as fundamental components of the Arcium Network and can be classified as Single Use, which execute one computation before being discarded, or Persistent, allowing for reuse. Clusters of nodes confirm their capacity to process tasks, perform necessary preprocessing, and retrieve inputs from the blockchain for concurrent processing. The network supports various MPC protocols, including BDOZ, to enhance efficiency and security.^{[5][7][8][9][10][11][12]}

Clusters

Clusters are designed as groups of Arx nodes that aim to utilize multiparty computation (MPC) for confidential operations. Customers can form Clusters based on various node properties, including reputation and computational capacity.

During the creation of a Cluster, customers specify parameters such as maximum load and the number of Trusted Execution Environment (TEE)-enabled nodes. Most properties are fixed, with the exception of potential increases in load.

Nodes must provide consent to join a Cluster. In non-permissioned Clusters, a randomly selected node is added, and activation requires approval from all nodes. A Node Priority List is maintained for backup nodes.

Clusters manage cryptographic key operations through Distributed Key Generation (DKG) and can employ TEEs for enhanced confidentiality. Cluster forking enables remaining nodes to conduct separate computations if an MXE is ejected.

Migration costs are shared among nodes, and intentions to shut down must be communicated to prevent penalties. Rewards are allocated based on stake delegation, with a Leader node collecting output shares for additional compensation. Permissioned Clusters enable organizations to oversee their node infrastructure.^{[13][14][15][16][17][18][19][20][21][22]}

Arx Node

Arx nodes function as fundamental components of the Arcium Network, aiming to facilitate distributed computing for complex tasks. Each node retains a single encrypted data fragment, employing the Multiparty Computation (MPC) protocol to enhance data privacy.

Nodes are associated with Node Operators, who may manage multiple nodes. Relevant metadata includes the node's IP, port, and jurisdiction codes. Arx nodes also manage key shares securely and may utilize Trusted Execution Environments (TEEs) to strengthen security.

Node Operators generate income through self-delegation and fees from third-party delegations. Reputation is evaluated based on historical performance, with the network currently supporting the BDOZ MPC protocol and plans for future protocol expansions.

Middlelayer Nodes aim to connect off-chain data to computations, requiring a degree of trust from users. Reliable infrastructure is essential for node operation, and the support of TEEs may increase potential revenue opportunities.^{[5][23][24][25][26][27][28][29][30][31][32]}

Computations

The Arcium Network differentiates between system and normal computations. System computations are designed to support network operations, including DKGSystemComputation for distributed key generation, MigrationSystemComputation for cluster migrations, and NonParticipationDetectionSystemComputation, which aims to trigger broader consensus when non-participation is detected.

Normal computations refer to customer-initiated tasks associated with an MXE, defined in Computation Definitions that specify outputs and execution authority levels. The execution logic utilizes public or private circuits based on the BDOZ protocol for efficient cryptographic operations. Customers are responsible for setting parameters for data sources while maintaining data integrity and confidentiality. A deterministic pricing model facilitates cost pre-assessment and allows for priority enhancements through additional fees, with node operators voting to establish sustainable pricing.^{[33][34][35][36][37][38][39][40]}

Decentralized Confidential Computing (DeCC)

Decentralized Confidential Computing (DeCC) aims to ensure secure processing of sensitive data without exposure. It combines decentralization, which distributes data across multiple locations, with confidentiality measures to protect against unauthorized access. This approach addresses growing concerns over data privacy while facilitating secure processing in the Web3 ecosystem.

DeCC utilizes technologies such as Multi-Party Computation (MPC), Zero-Knowledge Proofs (ZKPs), Fully Homomorphic Encryption (FHE), and Trusted Execution Environments (TEEs) to maintain data privacy during processing. The Arcium framework applies these DeCC principles to create a structure for encrypted computations, aiming to reduce the risk of data breaches by preventing any single entity from having full access to complete datasets.^[45]

Cerberus

Cerberus is a cryptographic protocol developed by Arcium for secure multiparty computation (MPC). It is built on the BDOZ MPC protocol and operates under the assumption of a dishonest majority, necessitating trust in only one honest participant per computation.

This protocol aims to incorporate an accountability mechanism through identifiable abort and oblivious transfer for trustless preprocessing. It allows for the identification and penalization of malicious actors in the network via a staking and slashing system based on cryptographic proofs.^[44]

Ecosystem

Solana Integration

Arcium utilizes Solana, an open-source blockchain that facilitates decentralized applications and transactions. On-chain programs on Solana are responsible for managing the Arcium Network, which includes the registration of Arx nodes, computation processing, and financial activities such as payments and rewards.

The network implements a dual-mempool structure: the active mempool for computations ready for execution and the dependent mempool for computations awaiting the completion of others. This design aims to improve organization by directing computations to clusters of Arx nodes. While the integration is initially with Solana, Arcium aims to support multiple blockchains in the future.^{[5][41][42][43]}

Use Cases

Arcium's confidential computing network aims to enhance data security and privacy across various sectors. Key use cases include:

• AI: Aims to facilitate the secure pooling of encrypted data for collaborative AI training while preserving the privacy of individual datasets.
• Decentralized Finance (DeFi): Aims to encrypt DeFi actions and expand into Real-World Assets (RWAs) to promote compliance and protect user data.
• Decentralized Physical Infrastructure Networks (DePIN): Aims to ensure secure handling of sensitive data in decentralized services without exposing the data.
• Healthcare: Aims to enable secure collaboration on sensitive patient data, facilitating analysis for personalized medicine while preserving privacy.^[5][44]