bitpro가 원문에서 글자만 모은 내용이에요. 표·그림·수식은 빠지거나 깨질 수 있어요. 전체 3쪽 중 1쪽
Injective: A Finance-Native Layer-1 Blockchain
Injective Research
October 2026
Abstract
Injective is a proof-of-stake Layer-1 blockchain designed for finance. It implements core
components of financial systems as protocol-native modules: a fully onchain orderbook with
front-running-resistant frequent batch-auction matching, oracle infrastructure, insurance funds
for derivative markets, real-world asset issuance and permissioning, and recurring token buy-
backs. Applicationscomposeagainstthesemodulesdirectly, sharingprotocolstateandaccessing
a common liquidity layer through the exchange module.
The network embeds the Ethereum Virtual Machine (EVM) and a WebAssembly (WASM)
virtual machine within one canonicalstate, unifiedbythe MultiVM Token Standard, andreaches
deterministic finality through Byzantine fault-tolerant (BFT) consensus. Injective-developed
modules provide the common financial layer, while EVM and WASM applications access it
without creating separate settlement domains. Deterministic module interfaces and machine-
payment pathways provide core agentic infrastructure, enabling users, institutions, and AI
agents to operate side by side against the same financial state. INJ is utilized to pay transaction
fees, secure the network through staking, and govern the protocol; it is also the asset acquired
and burned through the Community BuyBack.
This paper traces Injective’s evolution from an exchange settlement protocol into a sovereign
network and presents the resulting architecture as an integrated financial state machine. It
describes and formalizes the consensus threshold, block lifecycle, frequent batch auction clearing
rule, oracleaccounting, derivativeriskchecks, permissionmodel, anddynamicissuancecontroller
that govern the network’s operation. This document supersedes the original Injective Protocol
whitepaper of December 2018, which described a front-running-resistant exchange settlement
protocol on Ethereum.
This document supersedes the original Injective Protocol whitepaper (December 2018).
1
2 INJECTIVE WHITEPAPER OCTOBER 2026
1 Introduction
1.1 From settlement protocol to sovereign network
The original Injective Protocol whitepaper, published in December 2018, described a decentral-
ized exchange settlement protocol on Ethereum. Its design combined a trustless relayer network
for sharing liquidity with verifiable delay functions intended to resolve same-block order conflicts.
The protocol was framed by five properties: trustlessness, resolvability, public verifiability, liquidity
neutrality, and resistance to front-running.1
That architecture reflected the constraints of the time. Because the proposed protocol operated
above Ethereum, it could not directly determine the host chain’s consensus, transaction ordering, or
finality. It therefore needed additional machinery to establish fair ordering and coordinate exchange
settlement. The central insight, however, was durable: the quality of an onchain market depends
on the rules by which orders are admitted, sequenced, matched, and settled.
Injective subsequently developed into a sovereign Layer-1 blockchain, moving the objectives of
the 2018 design deeper into the stack rather than abandoning them. Consensus now supplies author-
itative ordering and deterministic finality, while the exchange module implements a decentralized,
fully onchain central limit orderbook (CLOB), applies matching and risk rules, and settles each mar-
ket transition. Liquidity comes from participants quoting into those markets, and any application
can reach the same books through the exchange module.
In making that transition, Injective structured the Layer-1 around modules: self-contained units
of state and execution logic that interoperate through defined interfaces. This architecture allowed
performance-sensitive financial primitives to become shared, validator-secured infrastructure and
evolve through targeted protocol upgrades. Around that market core, successive upgrades added
Injective-native modules for oracle integration, insurance infrastructure for derivative markets, buy-
back auctions, token issuance, asset-level permissioning, and scheduled contract execution. The
addition of a native EVM alongside WASM extended the system into a MultiVM architecture in
which different execution environments operate against one chain, one finality domain, and one
canonical asset state.2
This 2026 paper therefore does not introduce a replacement for the original protocol. It docu-
ments the live state machine that emerged from it. What began as an exchange settlement design
above another blockchain became a sovereign financial network able to define its own sequencing,
execution, risk, and settlement rules at the base layer.
1.2 Design thesis
Injective is organized around a single thesis: the common machinery of a financial system belongs
in the protocol layer. Four design principles follow from that thesis.
3 INJECTIVE WHITEPAPER OCTOBER 2026
Financial primitives and core onchain functions are implemented as shared protocol modules,
executed by validators as part of the chain’s state transition. This reduces duplicated infrastruc-
ture and places common financial rules under the network’s security and governance, rather than
requiring each application to rebuild and separately secure them in isolated contracts.
Markets require deterministic, final, and transparent execution. Financial positions and risk
controls depend on an authoritative order of actions, a common matching and risk model, and
settlement that does not expose participants to a probabilistic confirmation window.
One state can serve multiple execution environments. Developers, applications, and institutions
bring different technical requirements, and no single execution environment is optimal for every use
case. A MultiVM architecture allows complementary execution models to be used together without
fragmenting assets, liquidity, or settlement.
The token economy should evolve with the financial system it secures. Network security and
ecosystem growth place different demands on the native asset, so its economic design must support
both while remaining adaptable as participation and activity change over time.
Together, these principles define a Layer-1 network whose protocol layer supplies common fi-
nancial infrastructure while leaving its application layer open to developers, institutions, and au-
tonomous software.
1.3 Scope and implementation baseline
This paper describes Injective mainnet, chain identifierinjective-1, as of October 1, 2026.
Unless otherwise stated, descriptions of native modules, EVM, WASM, the MultiVM Token
Standard, exchange behavior, interoperability, governance, and token parameters refer to the soft-
ware and network state active on that date. Later software releases and governance actions may
change interfaces, parameters, or feature status. Source-code citations identify immutable commits;
claims about live behavior and active parameters identify a versioned release artifact or the relevant
governance record.2
2 Network architecture
2.1 Replicated financial state
Injective’s higher-level financial capabilities depend on a common systems requirement: every
validatormustmaintainthesamestateandapplythesameorderedtransitionstoit. Thisreplicated-
state foundation allows modules, virtual machines, and settlement paths to operate as parts of one
coherent network rather than as separate systems.
4 INJECTIVE WHITEPAPER OCTOBER 2026
At block heighth, Injective can be modeled as a replicated stateSh maintained by validators.
This state includes accounts and bank balances, bonded stake and validator voting power, and
governance proposals and parameters. It also records exchange deposits, orders, positions, and
market metadata, together with oracle observations, insurance-fund state, token and permission
namespaces, smart-contract state, and cross-chain client or bridge state. A valid block applies an
ordered transition function to the previous state.
Sh =T(S h−1, Bh)(1)
Equation 1. Canonical state transition
This formulation reduces a diverse financial system to one consensus requirement: every honest
validator computes the sameSh from the same prior state and block. Markets can be processed
in parallel where their state is independent, but all results are combined deterministically before
commitment.
The state transition is produced and used by several classes of participants:
•validators, which propose, execute, vote on, and commit blocks;
•delegators, which assign stake and voting power to validators;
•users and applications, which submit signed messages and smart-contract calls;
•oracle reporters and transmitters, which bring external observations on chain;
•relayers, bridge operators, and issuers, which connect external systems or assets;
•governance participants, which can change parameters and authorize upgrades;
•asset administrators, which may control permissioned namespaces; and
•agents and automated contracts, which act within delegated keys and policies.
2.2 Layered architecture
Injective organizes the system as a stack whose components converge upon one committed
state. Users, applications, institutions, and agents interact through native module messages, EVM,
or WASM. These interfaces invoke protocol state for accounts, assets, finance-specific modules, and
governance. Validators replicate that state and reach onchain consensus to agree on transaction
ordering and establish the canonical commit. Inter-Blockchain Communication (IBC), the Cross-
Chain Transfer Protocol (CCTP), Peggy, and other bridge infrastructure connect external domains
to the state machine through path-specific verification rather than forming sequential stages through
which every transaction passes.
5 INJECTIVE WHITEPAPER OCTOBER 2026
One canonical state, multiple execution and interoperability interfaces
Execution environments and verified cross-chain paths converge on one validator-secured state.
Users, institutions, applications, and Al agentsPARTICIPANTS
INJECTIVE NETWORK
CROSS-CHAIN INTEROPERABILITY
Path-specific verification connects external
domains to Injective state.
IBC
Light-client verified packets
EXECUTION AND ACCESS SIGNED TRANSACTIONS AND STATE QUERIES
Native module messages
Direct protocol interfaces
WASM
WebAssembly contract execution
EVM
Ethereum-compatible contract
execution
READ AND UPDATE ONE STATE CCTP
Circle-attested USDC burn and mint
CANONICAL STATE AND SETTLEMENT
Accounts, balances, and MTS
Canonical accounts , balances , and toke n
identity
In j ective financial modules
Exchange , oracle integration , insurance ,
and tokeni z ation
ONE SHARED LED G ER
G overnance and control
Staking , governance , parameters , an d
permissions
VERI F IE D
CROSS-CHAI N
TRANSITIONS Peggy and external bridges
Path-specific verification
SECURED AND F INALI Z ED BY
CONSENSUS AND F INALITY
Validator consensus
Proof of stake , proposal , prevote , and precommit
Canonical commit
M ore than t w o thirds of voting po w er establishes finality
Figure 1: Injective’s multi-layered architecture
The distinction between a module and a smart contract is central to this architecture. A contract
is application code deployed within an execution environment, whereas a module forms part of the
chain’s own transition function and is executed by every validator. By placing common financial
primitives in modules, Injective provides an onchain orderbook, risk state, balances, and settlement
logic as public infrastructure while leaving product-specific behavior to applications.
2.3 Accounts, signatures, and transactions
Injective accounts are controlled by public and private key pairs and use a native Bech32 address
beginning with inj. The same secp256k1 key can also be represented as a 20-byte Ethereum-style
address, giving one externally controlled account a deterministic native and EVM representation
rather than requiring a bridge between identities. An account number identifies the account, while
a sequence number advances with each accepted native transaction to prevent replay.2
At the native protocol interface, a transaction contains one or more module-specific messages
together with context such as the account number, sequence, chain identifier, fees, gas limit, memo,
and signatures. A message may transfer an asset, place an order, change a staking position, vote in
governance, or call another module function. Regardless of its contents, the transaction follows the
same preparation, signing, and broadcasting lifecycle before validators check it and include it in a
proposed block.2
6 INJECTIVE WHITEPAPER OCTOBER 2026
Native transactions and EVM transactions use different encodings and client interfaces, but
both enter the same chain and ultimately modify the same replicated state. Native signing paths
support module-oriented transaction formats, while EIP-712 allows Ethereum wallets to authorize
compatible Injective transactions using typed data. Across mixed native, WASM, and EVM oper-
ations, Injective’s MultiVM design preserves transaction atomicity so that a failed operation does
not leave a partially applied state transition.2,3
3 Consensus and deterministic finality
3.1 Consensus engine and commit rule
Injective uses proof of stake with an Injective-maintained implementation of CometBFT, the
successor to Tendermint Core. The current release branch explicitly replaces upstream CometBFT
with the Injective fork. The implementation preserves the Tendermint BFT proposal, prevote,
precommit, locking, and greater-than-two-thirds commit structure formalized below. LetWbe the
total bonded voting power,Fbe the voting power controlled by faulty validators, andCr(h)be the
precommit voting power for blockhin roundr. A block commits only when the precommit voting
power for that block exceeds two thirds ofW.2,4
Cr(h)> 2W3 , F < W3 (2)
Equation 2. Commit threshold and fault assumption
WithF < W/3, conflicting blocks cannot both commit. In one round, precommit sets above
2W/3overlap by more thanW/3and must include honest voting power; honest validators do not
precommit conflicting blocks. Locking and round changes extend safety across rounds; validators
advance past unavailable or invalid proposals under the same commit rule.4
3.2 Validator participation and accountability
Bonded INJ connects the consensus rule to economic participation. Validators bond their own
INJ and can receive delegated stake, which together determine stake-weighted voting power in the
active validator set. Active validators propose blocks, validate transactions and state transitions,
and exchange prevotes and precommits. Delegators contribute bonded stake to this security process
without operating consensus infrastructure.2
Bonded stake also provides the accountability layer. Protocol-defined faults, including extended
downtime and signing conflicting blocks at the same height, can trigger slashing and validator status
changes. Because delegated stake contributes to a validator’s voting power, those penalties affect
the stake bonded through that validator under the protocol’s rules. An unbonding period delays
the release of stake after participation so faults can be detected and attributed before the associated
7 INJECTIVE WHITEPAPER OCTOBER 2026
capital exits the security process. This section describes the consensus role of staking; Section 9
and the dedicated tokenomics paper address issuance and the broader reward economics.2,19
3.3 Consensus implementation and block production
Block latency is determined by more than the voting rule alone. Transactions must propagate
across the network, a proposal must reach validators, validators must execute and validate it, con-
sensus messages must circulate, and the resulting state must be persisted. Injective reduces latency
along this path through validator direct peering, optimized state synchronization and data handling,
more efficient disk and resource management, reduced I/O load, and refinements to consensus pro-
cessing. 2
These changes reduce latency along the block-production path without introducing a post-
commit confirmation period. Injective publishes a block time of approximately 600 milliseconds
and deterministic finality at commit. Block time and finality are distinct: block time describes
the cadence at which blocks are produced, while finality describes when a state transition becomes
irreversible under the consensus model.4,5
Short block intervals must still be balanced against block capacity and execution cost. Larger
proposals take longer to propagate and validate, while heavier application execution delays the
vote cycle. Block gas, mempool capacity, and consensus timing therefore form a coupled operating
envelope rather than independent maxima.2,5
3.4 Finality as a market property
A committed Injective block has deterministic finality: after commitment there is no longest-
chain fork choice and no probabilistic confirmation window.4
For finance, this distinction is substantive. A trade that appears in a block but may later
be reorganized creates credit and hedging uncertainty. Deterministic finality allows downstream
systems to treat the committed transition as authoritative immediately. A fill, position update,
token transfer, or liquidation is final with the block that contains it.
Finality also narrows the interval during which state can diverge across applications. Because
native modules and smart contracts settle against the same committed ledger, a transfer completed
through the EVM and a balance read through a native module converge on the same finalized result,
giving downstream systems a common state against which to account and manage risk.
8 INJECTIVE WHITEPAPER OCTOBER 2026
4 MultiVM execution, block construction, and shared
state
4.1 EVM and WASM
Injective embeds EVM and WASM execution within one Layer-1 state machine, alongside
purpose-built modules for exchange, oracle, insurance, token issuance, permissions, automation,
and other network functions. At the implementation layer, the chain combines Injective-developed
modules with foundational components built using the Cosmos software development kit (SDK); at
the developer layer, the relevant execution choices are EVM, WASM, and direct module messages.2
The EVM uses chain ID 1776 and belongs to the same Injective network identified natively as
injective-1; it is neither a separate rollup nor a bridged execution domain. Transactions invoking
Solidity and WASM contracts pass through the same validator consensus and reach finality with
the same canonical commit. Through EVM precompiles and WASM native-module messages, both
environments can update shared native module state. Where Solidity requires an EVM-native
interface to a module, Injective exposes that interface through a precompile, while WASM contracts
use native bindings and module messages. These are execution gateways within one chain rather
than merely interoperable bridges between separate ledgers.2
4.2 MultiVM Token Standard
The MultiVM Token Standard, or MTS, gives a token one canonical balance and identity across
execution environments. Rather than synchronizing an EVM token with a separate native copy, the
bank module maintains the authoritative balance and total supply, while EVM, WASM, and native
interfaces address that same underlying asset state.2
Two elements implement the EVM component of this model. The Bank precompile exposes
balance, supply, transfer, mint, burn, and metadata functions to Solidity, while the ERC-20 module
maps bank denominations to ERC-20 contracts. Existing TokenFactory, IBC, and Peggy denom-
inations can therefore acquire an EVM interface without becoming economically separate tokens.
ERC-20 allowances remain contract state because they are an EVM authorization concept, but
balances remain bank state, so a transfer from either environment mutates one canonical ledger
entry.2
Applications can use the same collateral and settlement assets across 프로젝트로 돌아가기