Research

Zero: Technical Positioning Paper

By MultipleFeb 10, 202613 min read

The technology underlying the Zero blockchain

Many blockchains today rely on so-called Layer 2 (L2) “solutions” to increase throughput. These L2s periodically post their data and execution proofs to a Layer 1 (L1) blockchain and claim to inherit the L1’s security. In practice, L2s have retained ultimate authority over their execution logic and can unilaterally modify it—typically through a security council composed of a small group of individuals.

However, upgradeability becomes a systemic vulnerability if controlled by any group that is not exactly equivalent to the L1 trust. In practice, these groups are small, resulting in a concentration of power that undermines decentralization and denies users the promises of permissionless, censorship-resistant blockchains. For all practical purposes, L2s are not scaling the L1, but are separate blockchains disconnected from the L1 in terms of trust and security.

Zero is a high-performance blockchain powered by multiple highly integrated shards, with control of execution logic retained by a single enshrined governance system at the settlement layer.

In this document, we start from the fundamental goal of blockchains—decentralization—and explain how Zero achieves it by allowing anyone to participate as a validator. This is made possible through a unified design that connects economics, governance, and system architecture. With a truly decentralized settlement layer, we enshrine a governance process that manages and upgrades execution logic across all shards, without granting unilateral control to any single entity or small group. We then explain how this governance process works and why Zero fundamentally differs from Layer 2-based approaches. Finally, we highlight our technical contributions that address the performance limits faced by blockchains today.

Decentralization-first: The ethos of blockchain

Decentralization means no single entity can gain unilateral control. It exists on a spectrum: the more difficult it is for any single actor or coordinated group to gain control, the more decentralized the system is.

We believe in decentralization through permissionlessness: participation in the network should be accessible to anyone and everyone. In centralized systems, transactions are often censored, contradicting the vision of a world computer. Decentralization underpins censorship resistance, ensuring end-users can always access the system and transact without permission. Decentralization is not just ideological—it is a hard technical requirement for these guarantees to hold credibly.

True decentralization is only possible with “nano validators”

True decentralization is only possible when anyone can permissionlessly run a validator, both in theory and in practice.

Most blockchains fail this test. Bitcoin home miners effectively cannot earn rewards without joining large pools. Ethereum validators must download over 1 terabyte of data and stake at least 32 ETH to participate. Solana validators require powerful machines and spend over 1 SOL every day just in voting fees.

Zero is designed to remove these barriers through a unified approach that aligns economic incentives, governance, and system architecture to support nano validators.

Economic decentralization: Eliminating small-staker traps

Zero prevents tokenomics-driven centralization through three economic designs that eliminate structural disadvantages for home stakers.

  • Pure Delegated Proof of Stake (PDPoS). Anyone can be a validator without providing a self-stake, and anyone can delegate stake without a minimum amount. We call this pure delegated proof of stake because validators do not need to rely on their own stake; all stake securing the network comes from delegation. This keeps participation open to all, regardless of capital size.
  • No consensus-layer slashing. Validators do not face catastrophic stake loss due to mistakes, downtime, or misconfiguration. Combined with PDPoS, this allows any token holder to delegate to any validator to support decentralization, without taking on slashing risk.

In practice, slashing is not an effective defense against sophisticated adversaries. If an attacker controls less than one-third of the total stake, an attack is guaranteed to fail due to the Byzantine fault tolerance of the consensus protocol. So the adversary would be imprudent to attack in the first place. If an attacker controls more than one-third of the stake (enough to violate safety), then they already have so much control over the system that they can prevent or bypass slashing.

  • Stable and proportional rewards. Staking rewards are distributed proportionally to all honest and online participants, independent of stake size. This avoids reward variance that forces small stakers to wait long periods before earning rewards.

Together, these designs ensure that every coin matters. Regardless of stake size, any staker can meaningfully contribute to the network’s security and decentralization by running a validator operator or delegating to one, strengthening decentralization.

Governance participation: Senators for non-expert home stakers

Decentralization is incomplete if governance is accessible only to full-time protocol experts.

In Zero, system changes—such as adding a new execution shard, modifying its state transition function, upgrading validator software, or adjusting protocol parameters—must go through an on-chain governance process voted on by all eligible stakeholders. However, home stakers may not have the time or knowledge to adequately review each of these proposals.

Zero solves this with a Senator model. Senators are individuals or entities recognized for their expertise in domains such as protocol design, cryptography, economics, or infrastructure. Stakeholders may delegate their voting power to Senators they trust, while retaining the ability to vote directly at any time.

This enables non-expert participants to meaningfully engage in governance without constant attention or deep technical specialization, delegating expertise without surrendering control.

Validity-based scaling removes the hardware barrier

In Zero, application execution happens in Atomicity Zones, asynchronously-executed shards of the Zero blockchain. Atomicity Zones are tightly integrated with and inherit the security of the settlement layer. Zones can be any type of computer application. User transactions are processed through the following steps:

  • Execution: Block Producers execute Atomicity Zone logic against user transactions.
  • Proof generation: Block Producers generate succinct validity proofs attesting to the correctness of the Atomicity Zone state transition execution.
  • Verification: Settlement layer validators verify the validity proofs, pack them into settlement layer blocks, and maintain consensus over the settlement ledger.

Proof verification is orders of magnitude cheaper than execution, dramatically reducing compute requirements of the consensus layer. 

As an optional optimization, Block Producers may batch user transactions into Atomicity Zone blocks. Crucially, the settlement layer retains final authority over Atomicity Zone block production, preserving censorship resistance. This design allows Block Producers to increase throughput without making them a single point of control for Atomicity Zones. 

Each Atomicity Zone Block Producer is given a defined time window to incorporate the block produced by the settlement layer. This allows honest and online Block Producers to provide low-latency preconfirmations to users. Blocks proposed by the Block Producer are ordered immediately, and the Block Producer may eagerly execute and order the block proposed by the settlement layer before the deadline. As a result, transactions from both the Block Producer and the settlement layer are confirmed quickly and in a deterministic order.

Zero introduces a designated Atomicity Zone, called the System Zone, which handles core functions often placed in the settlement layer on other blockchains:

  • Native coin (ZRO) circulation. The System Zone maintains ZRO balances and processes ZRO transfers.
  • Pure delegated proof of stake. The System Zone manages validator pool registration, stake delegation, and reward distribution, and periodically samples the consensus committee in proportion to stake.
  • Governance. Stakeholders participate in governance by submitting transactions to the System Zone.

Finally, the System Zone submits the required information to the settlement layer, including consensus committee membership and all settlement-layer parameters, such as updates to the execution logic of each Zone, so that this information can be verified.

By removing account management and governance from the settlement layer, validators maintain only a minimal consensus state, reducing storage requirements and allowing new participants to quickly join the consensus protocol.

Together, Zero’s economic, governance, and architectural designs enable home staker participation and make decentralization practical. This is how Zero enables nano validators.

The Noble Lie

Many blockchains today scale by using L2s, which execute transactions off-chain and periodically submit their data and execution results to an L1 blockchain, claiming to derive security from it. This approach relies on a core security property of blockchains: immutability. Ethereum is a classic example. Its L2 roadmap centers on rollups, which are implemented as smart contracts on Ethereum and perform transaction execution off-chain. Once a smart contract is deployed, its code is fixed and its execution logic is frozen.

In practice, the fundamental tension between immutability and upgradeability creates a dilemma:

  • Upgradeability. To remain competitive, applications must fix bugs, respond to new threats, and ship new features. This typically requires upgradeable contracts, most often implemented using proxy contracts controlled by an admin multisignature.
  • Immutability. Truly immutable systems cannot arbitrarily change their logic. In Ethereum, this can be achieved by deploying a smart contract with the full execution logic directly, rather than using proxy contracts with administrative control. However, by doing this, they also sacrifice flexibility: bugs become permanent, evolution becomes impossible, and obsolescence becomes inevitable.

This is the core L2 dilemma faced by all rollups deployed on Ethereum. 

The industry often treats this contradiction as a temporary compromise, assuming that centralized upgradeability is an acceptable step toward a more decentralized future. In practice, however, upgradeability is a permanent requirement for real-world applications. As long as the architecture of L1 blockchains does not fundamentally change, this tension between immutability and upgradeability cannot be fully resolved.

The protocol owns the zones

Fundamentally, applications fall into the dilemma because most L1s do not support upgradeability. As a result, critical issues in execution logic can remain unaddressed. 

Zero resolves this tension by uniformly managing upgradeability via protocol governance. The creation and modification of Atomicity Zones must go through a governance process that is open to all stakeholders and built directly into the protocol. Proposed upgrades are reviewed and scrutinized to ensure the state transition logic is correct and free from malicious behavior. Importantly, governance-based upgrades in Zero are not an emergency workaround—they are the intended and standard path for protocol evolution, designed to preserve trust while allowing the system to improve over time.

As a result, users do not need to fear silent or unauthorized upgrades. Changes are explicit, transparent, and collectively approved—preserving trust backed by all stakeholders while allowing the system to evolve.

This approach strikes a principled balance: immutability grounded in blockchain security, and upgradeability enforced through transparent, accountable governance. The result is a unified blockchain with an ultra-lightweight settlement layer and many execution shards, all sharing the same security foundation.

Unlocking the four bottlenecks of validity scaling

Zero’s architecture is designed to scale without centralizing. Our innovations in zkVM, data availability, state storage, and parallel execution enable Zones to deliver millions of verifiable transactions per second.

Jolt Pro: Proving Execution at Planetary Scale

Zero relies on Jolt, the fastest zero-knowledge virtual machine (zkVM) that can handle proving arbitrarily complex execution logic across different Atomicity Zones. Jolt is a breakthrough in zkVM design which is mathematically simpler and more efficient than other zkVMs. We collaborate with the Jolt team to make Jolt Pro.

Jolt Pro builds on Jolt with a state-of-the-art GPU-accelerated prover, a ZK-optimized implementation of EVM, and inlines. Inlines are an optimization technique that replaces high-level operations with sequences of RISC-V instructions; inlines achieve efficiency without introducing new ad hoc, error-prone constraints, unlike precompiles commonly used by other zkVMs. In Jolt Pro, we prove using cells, which are groups of colocated GPUs which collectively generate a proof. Our current flagship cell has 64 NVIDIA GeForce RTX 5090 GPUs and proves RISC-V at 1.61 billion cycles per second (1.61 GHz).

Jolt Pro currently implements succinct proofs, and zero-knowledge proofs are in the immediate roadmap.

SVID: solving the data withholding attack scalably

Under a naïve validity proof-based sharded blockchain, the Block Producer can withhold input data from the network, and potentially halt the Zone. The straightforward solution of sending the full block of data to every validator is not practical because it creates a network bottleneck.

We solve this problem with our data availability scheme: Scalable Verifiable Information Dispersal (SVID). In SVID, the Block Producer sends only a small slice of the block and a compact cryptographic fingerprint linking the slice to the original block. Validators collectively ensure that enough distinct slices are distributed across the network to prevent data loss. Our scheme is similar to that of Semi-AVID-PR [5]. SVID ensures that all non-faulty full nodes can remain synchronized to the most recent state, preventing Atomicity Zones from permanently losing liveness.

We tightly integrate the data flow between SVID and Jolt, allowing Jolt to prove the correctness of execution directly bound to the underlying data with almost no overhead. SVID achieves over 1 GB per second on nano validators today with a clear roadmap to achieve 10 GB per second as hardware performance scales.

QMDB: Solving the state storage bottleneck

Traditionally, blockchain state is stored in a verifiable data structure such as the Merkle Patricia Trie. These structures are fundamentally inefficient because the trie structure is persisted on disk, which requires random logical writes.

Our solution, the Quick Merkle Database (QMDB), extracts maximum performance from modern SSDs using a log-based append-only structure with verifiability designed in from the start. QMDB delivers 3 million state updates per second using a single machine, with one SSD read per state access, O(1) I/O for updates, and in-memory Merkleization with as low as 2.3 bytes of DRAM per entry. It can approach the performance of in-memory databases without their state size limitations or their high costs.

FAFO: Unleashing extreme parallelism in blocks

With QMDB, the execution bottleneck becomes compute. Most blockchains achieve determinism at the expense of parallelism by processing transactions serially. This limits the system from leveraging modern multi-core CPUs.

Our transaction scheduler, Fast Ahead-of-Formation Optimization (FAFO), reorders transactions prior to block formation to build highly parallelizable blocks while still providing determinism. FAFO uses Bloom filters to efficiently calculate data dependencies internally, in contrast to other blockchains that offload the problem onto dApp developers and offer high parallelism only if data dependencies are supplied. FAFO maintains compatibility with VMs (e.g., EVM) while executing up to more than 1.2 million EVM transactions per second with a single Block Producer on an AWS i8g.metal-24xl instance.

The Zero thesis

Decentralization is the core promise of blockchain technology. Without it, a blockchain degenerates into a centralized operator with needless overhead. True decentralization is achieved through permissionless participation by home validators. Zero’s minimal settlement layer enables nano validators, making decentralization operational and not just rhetorical.

At the same time, real systems must continuously evolve. Zero resolves the immutability–upgradeability dilemma through non-sovereign Zones, where execution logic can change safely without granting unilateral control or introducing hidden backdoors.

Zero unlocks every bottleneck in the stack to deliver performance without compromising principles. 

Zero is the decentralized world computer.

References

The LayerZero research team (authors are ordered alphabetically by last name: Bochun Bai, Andrii Baranov, Omid Bodaghi, Hao Chung, Quang Vu Dao, Qun Huang, Amirhossein Khajehpour, Thomas Kim, Mignon Li, Tan Li, Giuseppe Vitto, Daniel Lin-Kit Wong, Kelvin Wong, Saeid Yazdinejad, Kyle Zarick, Ryan Zarick, Isaac Zhang, Guangxian Zou.)

Thanks to Joachim Neu and Nusret Tas from a16z crypto research for their insightful feedback.

Connect to our team

[01]

Start building

Go to Docs
[02]

LAYERZERO 2026