Validium is a Layer 2 scaling architecture that combines zero-knowledge validity proofs (the security mechanism of ZK rollups) with off-chain data availability (storing transaction data outside Ethereum, unlike ZK rollups which post data to Ethereum mainnet). The distinction between a ZK rollup and a validium is precisely this data storage choice: both generate ZK validity proofs that mathematically prove the correctness of transaction execution; but a ZK rollup posts transaction data on-chain (on Ethereum, as calldata or blobs) ensuring anyone can independently verify and reconstruct state, while a validium keeps transaction data off-chain (typically with a Data Availability Committee: DAC — a trusted set of nodes that attest data is available) to dramatically reduce costs. This tradeoff delivers substantially higher throughput and lower fees (validiums can handle 10,000+ TPS vs. ZK rollup’s current practical 100-1,000 TPS on Ethereum DA) at the cost of a weaker security assumption: if the DAC becomes unavailable or colluding, users cannot prove their balances and may be locked from withdrawing funds from the validium. StarkEx (the technology powering dYdX v1, Immutable X, Sorare, and DeversiFi) pioneered the validium model in production. Immutable X — the leading blockchain for Web3 gaming — operates as a validium using StarkEx’s proof system with a permissioned data availability committee. The validium concept is generalized by Volition (a hybrid): users can choose per-transaction whether their data goes on-chain (ZK rollup mode) or off-chain (validium mode).
Key Facts
- Architecture: ZK validity proofs + off-chain data availability
- Key tradeoff: Higher performance / lower cost vs. weaker DA security
- Data Availability: Committee (DAC) or alternative DA layer (Celestia, Avail, EigenDA)
- Throughput: 10,000+ TPS (much higher than ZK rollup with Ethereum DA)
- Security assumption: Honest DAC + ZK proof system (two trust assumptions vs. ZK rollup’s one)
- Key implementations: StarkEx validium (dYdX v1, Immutable X, Sorare), Polygon validium chains (CDK + Avail DA)
- Use cases: Gaming, high-frequency trading, NFT minting — performance-sensitive applications
ZK Rollup vs. Validium: The Core Tradeoff
| Feature | ZK Rollup | Validium |
|---|---|---|
| Validity proof | ✅ ZK proof | ✅ ZK proof |
| Data on Ethereum | ✅ Always | ❌ Off-chain only |
| Data availability | Ethereum-grade (best) | DAC-grade (weaker) |
| Withdrawal if chain fails | ✅ Always possible | ❌ May be blocked if DA fails |
| Throughput | 100-2,000 TPS | 10,000-100,000 TPS |
| Cost per transaction | Higher (Ethereum DA cost) | Lower (no L1 data cost) |
| Use case | High-value finance | Gaming, trading, minting |
| Examples | Polygon zkEVM, Scroll | Immutable X, dYdX v1 |
Data Availability Committee (DAC)
What a DAC is:
- A trusted set of nodes (typically 5-20 members) that:
- Receive all off-chain transaction data before each batch is proven
- Sign attestation: “we have received and can provide this data”
- The validium’s on-chain contract: requires DAC signatures before accepting each batch
DAC security model:
- Assumption: At least 1 of N DAC members is honest and will provide data when requested
- If all DAC members go offline or collude: users cannot access transaction data → cannot reconstruct balances → cannot exit the validium
- Compare to rollup: Even if all network participants are offline, Ethereum still has the data → users can always exit
DAC member selection:
- Permissioned: DAC members are known, trusted entities (e.g., companies, foundations)
- StarkEx DAC: known entities committed to data availability
- Not decentralized (unlike Ethereum validators): known failure mode
Volition: The Hybrid
Volition (proposed by StarkWare) is a hybrid architecture:
- Per-transaction choice: user selects: ZK rollup mode (data on Ethereum; more expensive, more secure) OR validium mode (data off-chain; cheaper, slightly weaker security)
- Implementation: Same smart contract; different storage path based on user preference
- Use case: High-value DeFi trades → rollup mode; frequent low-value game actions → validium mode
- Current status: Theoretical/roadmap; not widely deployed in production
StarkEx: The Validium Pioneer
StarkEx (StarkWare’s L2-as-a-service product) powers several major applications:
- dYdX v1: DerivativesDEX on Ethereum (StarkEx validium; 2021-2023 before dYdX v4)
- Immutable X: NFT/gaming blockchain (StarkEx validium; 0-fee NFT minting)
- Sorare: Fantasy football NFT game (StarkEx validium)
- Rhino.fi (DeversiFi): Cross-chain trading (StarkEx validium)
Why StarkEx chose validium:
- Target applications: gaming + trading = performance-sensitive; not storing high-value DeFi positions
- Ethereum DA cost: prohibitive for NFT minting (would cost $5-50 per NFT in 2021-2022)
- Validium: enables free-to-mint NFTs (marginal DA cost: near zero)
- Security tradeoff: acceptable for gaming items; not acceptable for $1M DeFi positions
Alternative DA Layers
Beyond DACs, validiums can use alternative data availability layers:
| DA Layer | Security | Cost | Energy |
|---|---|---|---|
| Ethereum (calldata) | Highest | Highest | – |
| Ethereum (blobs/EIP-4844) | Highest | Lower (since March 2024) | – |
| Celestia | High (DAS-based) | Low | PoS |
| EigenDA | High (EigenLayer restaking) | Low | Restaking |
| Avail (Polygon) | High (DAS-based) | Low | PoS |
| Trusted DAC | Lowest | Lowest | Minimal |
Data Availability Sampling (DAS): Celestia, EigenDA, Avail use DAS — nodes only download small random samples of data; if samples are available, data is statistically available with high confidence. Stronger than a small DAC but weaker than full Ethereum replication.
Validium Security Risks
1. Data withholding attack:
- Malicious sequencer + DAC: withhold data after batch is proven
- Result: valid state proof on Ethereum, but no one can construct the state
- Mitigation: DAC signatures required before accepting batch; if DAC withholds, batch rejected
2. Data loss (not malicious):
- DAC goes offline unintentionally
- Result: valid state on chain; users cannot verify balances
- Mitigation: multiple DAC members with redundant storage; disaster recovery procedures
3. Regulatory:
- DAC members: known entities → vulnerable to government orders to withhold or censor data
Related Terms
Sources
- “Validium vs. ZK Rollup: When Is Off-Chain Data Availability the Right Tradeoff?” — Delphi Digital / Validium Architecture Research (2022). Foundational analysis of the validium vs. ZK rollup tradeoff — examining the specific scenarios where validium’s weaker security model is acceptable (gaming items, low-value NFTs, high-frequency trading with position limits) vs. where ZK rollup’s Ethereum-grade DA is required (DeFi protocols holding significant user value, stablecoins, lending positions). Analysis includes comparison of Ethereum DA costs (2022: extremely expensive → 2024: much cheaper post-EIP-4844) and how declining Ethereum DA costs narrow the cost advantage of validiums over time.
- “StarkEx Validium: How dYdX v1, Immutable X, and Sorare Achieved Scale Using Off-Chain Data” — Messari / StarkEx Research (2022-2023). Technical analysis of StarkEx as L2-as-a-service — examining how StarkWare deployed a configurable STARK-based validity proof system for multiple high-profile applications as validiums, the shared infrastructure benefits (one proving system; multiple applications), the DAC implementation (known entities; multi-sig attestation), and why StarkEx was the first production ZK-based L2 to achieve meaningful scale ($1B+ daily volumes on dYdX v1).
- “Data Availability: Why ‘Where Is the Data’ Is the Most Important Question in L2 Security” — Bankless / Data Availability Research (2023). Educational analysis of data availability (DA) as the fundamental security property of Layer 2 systems — explaining why “validity of computation” (what ZK proofs provide) is only half the security equation, and why “availability of data” (can anyone reconstruct the L2 state from publicly available information?) is the other half. Examining the spectrum of DA solutions (on-chain Ethereum calldata → EIP-4844 blobs → Celestia DAS → DAC → trusted centralized server) and their security/cost tradeoffs.
- “Immutable X: StarkEx Validium as the Foundation for Web3 Gaming” — DeFi Research Collective / Immutable X Analysis (2022-2023). Analysis of Immutable X’s role as the dominant Web3 gaming blockchain — examining how StarkEx validium enables gas-free NFT minting (the primary user experience requirement for gaming), the IMX token (governance + staking + fee payment), the Immutable Passport (embedded wallet for non-crypto gamers), and how Immutable X has signed partnerships with major game studios (Gods Unchained, Guild of Guardians, Illuvium) to position itself as the blockchain of choice for game publishers seeking a scalable, user-friendly NFT infrastructure.
- “Validiums, Volitions, and the Future of Off-Chain Data Availability in Ethereum Scaling” — Ethereum Research / Data Availability Roadmap (2023). Forward-looking analysis of how validiums fit into Ethereum’s long-term scaling roadmap — examining how EIP-4844 (reducing Ethereum DA costs by 10-40x), Danksharding (roadmap to further 64x DA capacity increase), and alternative DA layers (Celestia, EigenDA) reshape the validium vs. rollup tradeoff. Assessment of whether pure Ethereum rollups will eventually have sufficient capacity and low enough cost to make validiums unnecessary, or whether off-chain DA will always serve a cost/performance niche.