State of Monad: Architecture, Ecosystem & Open Questions

Published: March 17, 2026

tl;dr

• Monad targets 10,000 TPS with 400ms blocks and 800ms finality — built from scratch in C++/Rust by former Jump Trading engineers, using four interlocking innovations: MonadBFT, deferred execution, parallel execution, and MonadDb.
• Mainnet launched November 24, 2025 with ~$186–347M TVL — but daily active addresses dropped 92% from the TGE peak (150K to ~12K) and ~75% of bridged assets have flowed back out.
• ~90% of TVL sits in ported protocols, not native apps — Uniswap V3 (~$28M), Gearbox (~$20M), and other blue-chip deployments dominate; native Monad applications are struggling to capture share.
• $522M raised total including a $269M public token sale — MON trades at ~$0.022 (79% below ATH), with FDV of ~$2.2B; insider unlocks begin November 2026 (~46.5% of total supply).
• Governance is early-stage and opaque — the Foundation controls 38.5% of supply (unlocked), no verifiable on-chain votes exist, and MON’s official tokenomics page does not list governance as a token function.
• Key structural risks remain unaddressed — no public state growth data, single client implementation, no “only on Monad” application identified, and Ethereum’s Glamsterdam upgrade narrows the throughput gap.

Table of Contents

1. Executive Summary
2. Founding Team & Corporate Structure
3. Technical Architecture
4. EVM Compatibility
5. Network Status & On-Chain Metrics
6. MON Token & Funding History
7. Ecosystem & DeFi Landscape
8. MEV Dynamics Under Deferred Execution
9. Governance
10. Open Questions & Structural Risks
11. Sources

1. Executive Summary

Monad is a high-performance Layer 1 blockchain that maintains full EVM bytecode compatibility while targeting 10,000 TPS, 400ms block times, and 800ms finality. Built from the ground up in C++ and Rust by former Jump Trading engineers, the chain achieves its performance through four interlocking architectural innovations: MonadBFT consensus, deferred (asynchronous) execution, optimistic parallel execution, and MonadDb — a custom state database with async I/O that bypasses the filesystem entirely.

Mainnet launched November 24, 2025. As of March 2026, the network has 200+ validators, approximately $186–347M in TVL (sources diverge), and generates ~$15K/day in chain fees. The project raised approximately $522M total including a $269M public token sale on Coinbase. MON trades at ~$0.022 with a fully diluted valuation of ~$2.2B.

The chain is at an inflection point: technical architecture is genuinely differentiated, but ecosystem traction is heavily subsidy-driven (~90% of TVL from ported protocols, ~75% of bridged assets have flowed back out, daily active addresses dropped 92% from the TGE peak). Governance is early-stage and opaque. The competitive landscape is tightening as Ethereum’s Glamsterdam upgrade targets parallel execution on L1.

2. Founding Team & Corporate Structure

Founders

Keone Hon — Co-Founder and CEO

• MIT — B.S. in Computer Science and Mathematics, M.Eng in CS, Master of Finance
• 8 years at Jump Trading leading a high-frequency trading team; later transitioned to Jump Crypto
• His experience building low-latency, high-throughput trading systems directly informed Monad’s architecture

James Hunsaker — Co-Founder and CTO

• University of Iowa — B.S. in Computer Science and Mathematics
• Associate at Goldman Sachs, VP at J.P. Morgan, VP at Goldman Sachs. Joined Jump Trading in 2014 as senior software engineer — 8 years leading development of an ultra-low-latency trading system handling tens of billions in daily notional in major futures markets

Eunice Giarta — Co-Founder and COO

• MIT — S.B. in Computer Science and Engineering, minor in Management Science and Finance
• Background spanning Goldman Sachs/BofA Merrill Lynch (rates derivatives), Broadway Technology (senior PM), Shutterstock (senior PM), Jean-Georges Management (pastry chef), and entrepreneurship

Corporate Structure

Monad Labs rebranded into two entities in early 2025:

The Foundation was established December 2024, announced February 2025. It controls 38.5% of total MON supply (38.5B tokens) designated for “Ecosystem Development” — unlocked at launch. Category Labs holds an additional ~3.95B tokens (4-year vesting with 1-year cliff).

The codebase is licensed under GPL-3.0.

3. Technical Architecture

Monad’s performance comes from four interlocking innovations, each addressing a different bottleneck in traditional EVM execution.

3.1 MonadBFT Consensus

MonadBFT is a custom Byzantine Fault Tolerant consensus protocol building on the HotStuff family with critical improvements.

The Tail-Forking Solution: In pipelined BFT protocols (like HotStuff), a malicious leader can “fork away” the predecessor’s block to extract MEV. MonadBFT introduces the No-Endorsement Certificate (NEC): a leader must repropose the previous leader’s block unless they can obtain 2f+1 validators attesting they did NOT see that block. This prevents malicious forks while allowing legitimate leader replacement when the prior leader genuinely failed.

MonadBFT has been published as a formal research paper on arXiv and reviewed by the Stanford Blockchain Review.

3.2 Deferred (Asynchronous) Execution

This is Monad’s most consequential architectural innovation. It fundamentally decouples consensus from execution.

The problem: In Ethereum, the leader must execute all transactions, compute the state root, include it in the block, then validators re-execute to verify. Execution is constrained to ~1% of the 12-second block time (~100ms).

Monad’s approach:

• Nodes reach consensus on transaction ordering without executing any transactions
• The leader proposes an ordering without knowing the resultant state root
• Validators vote on block validity without verifying transaction execution
• Execution happens asynchronously after consensus, in a separate pipeline

Delayed Merkle Root: Since state roots aren’t available at consensus time, Monad includes a Merkle root from D blocks ago (currently D=3, ~1.2 seconds). This provides validation checks and divergence detection without blocking consensus.

Trade-offs:

• Reserve balance rules ensure transactions’ gas costs can be covered at consensus time despite the execution lag
• Newly-funded accounts must wait ~D blocks (~1.2s) before spending received tokens
• Stale state for MEV participants — builders cannot simulate against current state (see Section 8)

3.3 Parallel Execution

Monad executes transactions concurrently using optimistic concurrency control:

1. Speculative parallel phase: Transactions dispatched across multiple threads simultaneously, assuming no conflicts
2. Conflict detection: System tracks inputs consumed during execution and compares against outputs of prior transactions (in block ordering). If an input was modified by an earlier transaction, conflict detected.
3. Re-execution: Conflicting transactions re-executed with correct data. Signature recovery is cached; only state-dependent portions recomputed.
4. Sequential state merging: Updated state merged in original transaction order to preserve Ethereum’s deterministic execution semantics.

This is transparent to developers — no annotations, parallelism hints, or code modifications needed. The final state is identical to what sequential execution would produce.

The system relies on the empirical observation that most transactions touch disjoint state. Conflicts are rare in practice, making the optimistic approach highly efficient.

Not disclosed: Exact thread count, scheduling algorithms, or input/output tracking data structures.

3.4 Pipelining

Rather than processing blocks sequentially (consensus on N, execute N, consensus on N+1), Monad pipelines stages:

Execution occupies the full block time (400ms) in its own pipeline, not a fraction of it. CPU cores stay busy with consensus and execution running concurrently.

3.5 MonadDb

MonadDb is a custom key-value state database purpose-built for blockchain execution, replacing the generic databases (LevelDB, RocksDB, LMDB) used by most Ethereum clients.

Key design choices:

Explicitly NOT memory-mapped: MonadDb avoids mmap because “memory-mapped storage is implemented by the kernel and is not asynchronous, so execution is blocked while waiting for the operation to complete.”

Not available: No public benchmarks (IOPS, latency percentiles, throughput). No comparisons to LevelDB/RocksDB/MDBX under equivalent loads. No performance degradation data as state grows.

3.6 RaptorCast Block Propagation

RaptorCast distributes block proposals using erasure coding (Raptor codes, RFC 5053 with Monad modifications) through a two-level broadcast tree:

1. Leader sends erasure-coded chunks to non-leader nodes (proportional to stake weight)
2. Each non-leader relays assigned chunks to all other validators
3. Original block reconstructable from any sufficiently large chunk subset
4. Propagation latency bounded by RTT between the two most distant nodes
5. With 2x replication, reconstruction succeeds even if 33% of chunks fail (matching BFT assumption)

4. EVM Compatibility

Monad provides bytecode-level EVM compatibility — the strongest form:

• Identical EVM opcode execution; compiled bytecode runs without modification
• All opcodes and precompiles supported as of Ethereum Cancun fork
• Standard tooling works natively: MetaMask, Hardhat, Foundry, Remix, Ethers.js, Web3.js
• Full Ethereum JSON-RPC API compatibility
• Existing audited contracts (Uniswap, Aave, Curve) deploy as-is
• JIT compilation of EVM bytecode to native code (mentioned but not detailed)

The Monad team ran extensive simulations of historical Ethereum transactions and confirmed identical results to geth.

Known Differences from Ethereum

These are real architectural differences that could affect applications:

5. Network Status & On-Chain Metrics

Mainnet

Important distinction: The ~$168M in daily DEX volume is NOT fee revenue. Daily chain fees are ~$15K — a tiny fraction of volume. For context, Ethereum generates $1–5M+ in daily fees.

Staking

The 12.3% staking rate is extremely low compared to mature PoS chains (Cosmos ~65%, Solana ~67%, Ethereum ~28%). This likely reflects the network’s youth (~4 months old) and that only ~11% of total supply is circulating.

Validator Economics

Foundation dependency: Most validators rely on the Foundation’s Validator Delegation Program (~15B MON initially, planned 15–25B in year one). Without Foundation delegation, few validators could meet the 10M MON minimum total stake requirement.

Testnet History

• February 19, 2025: Testnet launch
• March 2025: 650M+ transactions, 72M active addresses
• May 2025: 1.55B transactions, 99 validators across 19 countries
• July 2025: 2.3B+ total transactions

6. MON Token & Funding History

Funding

The Series A was reported by PitchBook as the largest crypto funding round of 2024. Notable angels: Inversebrah, Ansem, Hsaka, punk6529, Eric Wall, Rune Christensen, Bryan Pellegrino, Luca Netz, Mert Mumtaz.

Token Distribution

Current circulating supply: ~10.83B MON (~10.8% of total). Insider unlocks begin November 24, 2026 — a major supply overhang.

Token Metrics (March 2026)

7. Ecosystem & DeFi Landscape

Protocol Breakdown

Critical observation: ~90% of TVL sits in established ported protocols (Uniswap, Curve, Morpho, Gearbox, Upshift) rather than native Monad applications. Native apps are struggling to capture market share against blue-chip DeFi deployments.

Bootstrapping: Subsidized vs. Organic

The data strongly suggests the majority of TVL is incentive-driven:

Key Ecosystem Developments

Aave v3 Deployment: TEMP CHECK passed on Aave governance (873K+ votes in favor, zero against). Monad Foundation committed $15M in incentives plus purchasing 10M GHO. Deployment possible mid-to-late March 2026.

Chainlink cbBTC Bridge: Launched March 2, 2026 via Chainlink CCIP, enabling cbBTC bridging from Base to Monad. The $5B figure in media reports represents total cbBTC circulation, NOT actual bridged amounts — the bridge enabled the capability, not a $5B transfer.

Kuru Exchange: Hybrid CLOB-AMM DEX built natively on Monad. Raised $13.6M total ($2M seed, $11.6M Series A led by Paradigm). Live on mainnet.

Oracle Infrastructure: Pyth (standard pull oracle, 400ms updates; Pyth Lazer for 1ms updates), RedStone, Supra (sub-penny feeds, 100ms latency), Chainlink (via CCIP).

The “Only on Monad” App Question

No application has been identified that structurally requires Monad’s specific combination of capabilities and could not exist on any other chain. Monad’s own documentation states that “the result of executing transactions in a block is identical between Monad and Ethereum.” The parallelism is an optimization, not a new programming model. Monad enables faster/cheaper versions of existing apps, not structurally new ones.

AI & Agent Ecosystem

• Monad AI Blueprint — support program for AI teams (partners: Google Cloud, Aethir, IO.NET)
• 742+ projects deployed across 2026 hackathons; 55.6% described as “agentic”
• Significant narrative around AI on Monad, but no public data on actual agent-driven transaction volume or economic impact

8. MEV Dynamics Under Deferred Execution

Monad’s deferred execution creates a fundamentally different MEV environment from Ethereum. This section draws primarily from aPriori’s published analysis — the most detailed public treatment of this topic.

How Deferred Execution Changes MEV

Because consensus finalizes transaction ordering before execution happens (potentially lagging by 1–2 blocks), the “current” state at block-building time may be N-1 or N-2 blocks behind the execution frontier.

Implications for MEV participants:

• Full block simulation within operational time windows is impractical
• Builders must infer searcher tip viability via reputation or simulation against stale (N-2) state
• Searchers face greater execution uncertainty — no theoretical guarantee against reverts
• Higher risk of bundles reverting on-chain vs. Ethereum’s MEV-Boost model

Local Mempool Design

The combination of unencrypted transactions, known leader schedule, and gas-priority ordering creates a familiar MEV attack surface. Leaders and observers can see transaction contents before inclusion.

MEV Infrastructure: Two Major Players

aPriori (MEVA + aprMON + Swapr)

aPriori operates a Monad Execution Value Auction (MEVA):

1. Searchers submit bundles
2. Builders construct partial blocks (top-of-block only) from searcher packages
3. Validators append remaining transactions from local mempool
4. MEV revenue redistributed to stakers via aprMON liquid staking token

Key innovation — Probabilistic Valuation: Because full state simulation is impractical, builders use probabilistic checking — estimating bundle success likelihood based on reputation, partial simulation against stale state, and statistical models. Explicitly compared to HFT strategies operating under similar uncertainty.

Swapr: AI-powered DEX aggregator classifying trades in <500ms using XGBoost, LightGBM, RNN, and Transformer models. Routes “clean” organic orders to efficient pools; isolates toxic flow. Swapr v2 planned with predictive routing and private execution.

Funding: $2M seed (2023), $8M (Pantera/Consensys/Flow Traders, July 2024), $20M (Pantera/HashKey/IMC Trading, August 2025). Total: $30M.

FastLane / shMonad (shMON)

shMON is a liquid staking token built by FastLane Labs:

• Users deposit MON, receive shMON
• Revenue from: native staking rewards + MEV + execution layer income
• Dual-deposit model: yield-focused mode or “Degen Depository” (higher reward multipliers)

Chainlink acquisition: January 22, 2026, Chainlink acquired Atlas (FastLane’s transaction ordering solution). FastLane continues operating independently; Atlas technology moves under Chainlink for cross-chain SVR (Smart Value Recapture). SVR has processed $460M+ in liquidations and recaptured $10M+ in OEV.

MEV Transparency: A Major Gap

No dedicated Monad MEV analytics exist. No Dune dashboards for Monad MEV, no published sandwich attack data, no arbitrage extraction data, no academic research, no Flashbots coverage. For a chain 4 months old with hundreds of millions in TVL, this absence is notable. Both aPriori and FastLane are operating MEV infrastructure but neither has published transparency reports.

Protocol-Level MEV Protections

Monad relies on out-of-protocol MEV infrastructure (aPriori, FastLane) rather than building PBS or MEV mitigation into the protocol. The local mempool (3-leader forwarding) provides some natural mitigation vs. a global mempool, but ordering is gas-priority and unencrypted — a classic MEV-susceptible configuration.

9. Governance

Current State: Early-Stage and Opaque

Monad’s governance structure is early-stage with a significant gap between what is publicly claimed and what is concretely documented.

What exists:

• Monad Research Forum (forum.monad.xyz) hosting MIP discussions
• MIP framework modeled after Ethereum’s EIP process (MIP-1 still in Draft status)
• GitHub repository (monad-crypto/MIPs) for specifications
• Active MIPs: MIP-1 (guidelines), MIP-3 (linear memory cost), MIP-4 (reserve balance introspection), MIP-5 (Fusaka EIP activation), MIP-6 (MONAD_NINE meta), MIP-7 (extension opcodes)

What does NOT exist (or cannot be verified):

• No verifiable evidence of on-chain governance votes having been executed
• No Snapshot space, Tally page, or other voting platform publicly identified
• No governance smart contract addresses, quorum thresholds, or voting period documentation
• MON’s official tokenomics page lists only gas fees and staking — governance is NOT listed as a token function

The MONAD_NINE Upgrade (Case Study)

The first named network upgrade bundled MIP-3 through MIP-5 into MIP-6. The upgrade was “activated by the development team” — no formal token vote or validator vote was documented. Forum discussions occurred (MIP-3 received 14 replies), but the approval process resembles Ethereum’s rough consensus model rather than formal on-chain governance.

Power Distribution

The Foundation’s 38.5% unlocked allocation would give it overwhelming voting power in any token-weighted governance system.

Comparison to Other L1s

10. Open Questions & Structural Risks

State Growth at 10,000 TPS

No public data exists on Monad’s actual state size or growth rate. This is the central sustainability question for a high-TPS chain.

• Ethereum grows ~50–100 GB of state/year at ~15 TPS
• Naive linear scaling to 10,000 TPS: 33–66 TB/year (obviously unsustainable)
• Monad’s 2TB SSD requirement suggests the team expects pruning to keep state manageable
• No state expiry plans, state rent, or Verkle tree migration have been discussed publicly
• MonadDb’s approach relies entirely on pruning historical state — it does NOT address current state growth (accounts, storage slots, contract code)

Client Diversity: Single Point of Failure

Monad has one client implementation (C++ execution + Rust consensus, both by Category Labs). No second client exists, is announced, or is in development. The deeply custom architecture (MonadDb, parallel execution engine, bare-metal requirement) makes alternative implementations exceptionally difficult.

This represents a meaningful centralization risk — a bug in the single client could take down the entire network. For context, Ethereum’s client diversity took years and significant dedicated funding to achieve.

The Vanguard Claim

Keone Hon has described Monad as a “vanguard environment” testing improvements that can flow back to Ethereum. Evidence of concrete upstream contributions: zero.

• No EIPs authored or co-authored by Monad team members
• No contributions to geth, Reth, or other Ethereum clients
• No research papers presented to the Ethereum research community
• Ethereum’s parallel execution work (EIP-7928 for Glamsterdam) appears independently developed

The claim is aspirational. Monad’s open-source code could theoretically be studied by Ethereum teams, but no evidence this has occurred.

Glamsterdam: The Competitive Threat

Ethereum’s Glamsterdam upgrade (planned H1 2026) targets parallel execution on L1 via EIP-7928 (Block-Level Access Lists). Testing shows ~78% of full parallel throughput achievable (10.8–13.9 GGas/s on 16-core machines).

Monad retains advantages even post-Glamsterdam:

• Sub-second finality (Ethereum unlikely to match soon)
• Purpose-built database (MonadDb) vs. generic Ethereum storage
• Asynchronous execution (not just parallel)
• Deeply integrated architecture built from scratch

But Glamsterdam narrows the gap:

• “1000x faster than Ethereum” narrative weakens if Ethereum gets 5–10x faster
• Ethereum L2s with parallel execution could match or exceed Monad throughput
• Monad’s remaining advantage shifts from raw throughput to fees and finality

The team has not publicly addressed this competitive scenario.

Summary Risk Matrix

Data Sources & Methodology

Data Sources: Monad documentation (docs.monad.xyz), DeFiLlama, CoinGecko, CoinMarketCap, official blog posts, third-party research (Blockworks, Stanford Blockchain Review, Chorus.one, Figment), protocol documentation

Methodology: All quantitative data sourced from public dashboards and verified documentation. Where specific data points could not be independently verified, limitations are noted. No data has been fabricated. Fee revenue and trading volume are clearly distinguished throughout.

11. Sources

Official Documentation

• Monad Homepage
• Monad Developer Documentation
• MonadBFT Documentation
• Parallel Execution Documentation
• Asynchronous Execution Documentation
• MonadDb Documentation
• RaptorCast Documentation
• Local Mempool Documentation
• Staking Documentation
• Hardware Requirements
• Gas Pricing
• MON Tokenomics Overview
• Monad Blog — How Monad Works
• Monad Blog — Home of High Frequency Finance
• Monad Blog — Next Gen CLOB
• Introducing Monad Foundation
• Introducing Category Labs

Research Papers

• MonadBFT: Fast, Responsive, Fork-Resistant Streamlined Consensus (arXiv)

Team Profiles

• Keone Hon — IQ.wiki
• James Hunsaker — IQ.wiki
• Eunice Giarta — IQ.wiki

Third-Party Research & Analysis

• Stanford Blockchain Review — Unpacking MonadBFT
• Blockworks Research — Monad: A Supercharged EVM L1
• Figment — Monad First Look
• Chorus.one — Deep Dive into Monad’s Architecture
• DeSpread — Monad: A New Paradigm in Community Building
• Blockworks — After Monad’s TGE

MEV & Infrastructure Sources

• aPriori — MEV Landscape in the Parallel Execution Era
• aPriori — Introducing Swapr
• aPriori — Go With The Flow
• aPriori Docs
• shMonad Docs
• Chainlink Acquires Atlas (PR Newswire)
• Magma — MEV North Star
• Kuru Docs

Governance Sources

• Monad Research Forum
• MIP-1: MIP Purpose and Guidelines
• MIPs GitHub Repository
• Monad Initial Specification Proposal (PDF)

Ecosystem & Market Sources

• Uniswap Blog — Monad Mainnet Live
• Yellow News — Native Monad Apps Struggle
• MEXC — Monad Ecosystem Update After TGE
• Monad Blog — cbBTC to Monad
• Aave Governance — Deploy on Monad TEMP CHECK

Funding Coverage

• The Block — Monad Labs raises $225M
• Fortune — Paradigm leads $225M round
• PitchBook — Monad Labs unicorn status
• CoinCentral — $269M Token Sale

Market Data

• DeFiLlama — Monad
• DeFiLlama — Monad Fees
• CoinGecko — MON
• CoinMarketCap — MON

Competitive Context

• EIP-7928 — Block-Level Access Lists
• Ethereum Glamsterdam Upgrade (CoinTelegraph)
• Bankless — Inside Monad’s Parallel EVM with Keone Hon

Data Limitations

1. TPS verification: The 10,000 TPS claim is from Monad’s documentation. No independent third-party benchmark confirms sustained 10,000 TPS under real mainnet load.
2. State growth: No public measurements of actual state size or growth rate exist. This is the most significant data gap.
3. TVL discrepancy: Sources report $186M to $347M, reflecting different dates and methodologies.
4. MEV data: No public MEV analytics exist for Monad mainnet despite active MEV infrastructure.
5. MonadDb benchmarks: No IOPS, latency, or throughput figures have been published.
6. Governance voting: No verifiable on-chain governance votes were found despite third-party claims.
7. MON ATH date: Reported as October 10, 2025 (before mainnet launch) — may reflect pre-market trading.
8. Thread count and internal parallelism parameters: Not disclosed in public documentation.

No data was fabricated in this report. Where specific numbers could not be independently verified or showed discrepancies across sources, this is explicitly noted.