Market cap $2.18 T −3.99 % over 24 hours24h volume $70.38 BBTC dominance 59.3 %ETH 10.1 %Fear & Greed 10 Extreme fearCoins 22,053

Project · Ethereum Layer 2

MegaETH (MEGA): a 10-millisecond Ethereum layer 2 with one sequencer, instant upgrades and a 96% fall in TVL

Active Risk 4/5

About the project

MegaETH is an Ethereum layer 2 built to answer in milliseconds: one sequencer orders every transaction and seals a “mini-block” about every 10 milliseconds. Its MEGA token launched on 30 April 2026. Six and a half months later the chain holds a fraction of what it attracted at launch. DefiLlama counted $8.81 million of total value locked (TVL) on MegaETH on 15 November 2026, against a peak of $245.6 million on 12 May.

The team behind it has changed course. On 10 November 2026 the MegaETH Foundation renamed itself MEGA and pledged $30 million from its treasury to new products, the first an AI inference market planned for December. Three days later co-founder Shuyao Kong said MEGA had cut about 20% of its staff, as our report on the MEGA layoffs explained. The chain keeps the MegaETH name; the token was already called MEGA.

What is MegaETH? A real-time layer 2 on the OP Stack

The network is a layer 2 network that settles on Ethereum. It runs on Optimism’s OP Stack, the open-source framework behind Base and OP Mainnet, under chain ID 4326. Gas is paid in ETH, assets arrive from Ethereum through a canonical bridge, and ordinary EVM wallets and contracts work on it.

The difference is the clock. OP Mainnet and Base seal a block every two seconds. MegaETH’s documentation describes a chain “optimized for real-time execution”, aimed at apps where a second is too long: order books, on-chain games, one-tap prediction products and payments. Its homepage advertises more than 100,000 transactions per second, more than 10 billion gas per second and block times under 10 ms. Those are the project’s own numbers; the section on speed below sets out what has been measured.

MegaLabs, the developer, started raising its seed round in December 2023 and announced it in June 2024. The public mainnet opened on 9 February 2026, and the token followed once the first network target was met in April.

How MegaETH works: one sequencer and mini-blocks every 10 ms

The speed comes from a deliberate choice. Instead of spreading block production across many machines, MegaETH relies on one sequencer, which the docs call “the single node responsible for ordering and executing transactions”. It executes transactions as they arrive and streams the results out. A standby node is kept ready and, according to the documentation, takes over within tens of milliseconds if the active one fails, so upgrades do not pause the chain.

Mini-blocks and EVM blocks

The sequencer produces two kinds of blocks. About every 10 milliseconds it seals a mini-block: transactions, receipts and state changes under a compact header with no state root, no bloom filter and a timestamp precise to the microsecond. About every second it seals a standard EVM block covering the mini-blocks of that second. Each transaction sits in exactly one of each.

Block explorers and most wallets read the one-second EVM blocks. Apps that want the 10 ms view subscribe to mini-blocks through MegaETH’s Realtime API over WebSocket. Since the Rex5 upgrade the sequencer signs each mini-block header with its sequencer key, a signature anyone can check on-chain. The docs are careful about what that proves: the sequencer committed to the contents, which is not the same as finality.

Each mini-block is streamed to RPC nodes in several regions. Replica nodes, the default, apply the results without re-executing them. Full nodes re-execute every block so they do not have to trust the sequencer, and a stateless validator, which the docs say runs on commodity hardware, checks each block against cryptographic witnesses.

Data on EigenDA, disputes through Kailua

MegaETH does not post its full transaction data to Ethereum. The docs name EigenDA as the primary data availability layer: block data goes to EigenDA, which returns a certificate, and without that certificate the sequencer cannot submit a block to Ethereum. L2BEAT, which rates layer 2 security, writes that the sequencer falls back to Ethereum if EigenDA is unavailable. Because data sits outside Ethereum and challengers are few, L2BEAT lists MegaETH among “Others” rather than with rollups.

Disputes over the chain’s state go through Kailua, a fault proof system built on RISC Zero’s zero-knowledge virtual machine. A challenger settles a dispute with one proof instead of the many rounds of the standard OP Stack game. The design is still optimistic: a state root counts as valid unless challenged within seven days, followed by an execution delay of three and a half days, according to L2BEAT.

Execution runs on MegaEVM, an EVM implementation built on the Rust libraries revm and op-revm. State is kept in SALT (Small Authentication Large Trie), an authenticated key-value store whose authors say its authentication layer needs about 1 GB of memory for up to 3 billion entries, so a node avoids random disk reads. Gas comes in two parts, compute gas and storage gas, both drawn from the transaction’s single gas limit, so writing to state is priced apart from running code.

Speed claims: what is measured and what is marketing

No independent measurement of the 100,000 TPS homepage figure has been published. When the public mainnet opened on 9 February 2026, The Block reported targets of 50,000 TPS and 10 ms blocks. In January MegaETH had announced a seven-day stress test aiming at 11 billion transactions and a sustained 15,000 to 35,000 TPS, driven by games and by ETH transfers and swaps the team ran itself on the Kumbaya exchange. A test of that kind shows what one powerful sequencer can process. It says nothing about whether real users will bring that load.

Stacked blocks, a cloud and a gear wheel linked by a web of nodes
One sequencer seals a mini-block about every 10 ms and streams it to RPC nodes, while block data goes to EigenDA.

Who controls MegaETH: one proposer, a 6-of-10 multisig, no exit window

The speed has a governance cost. L2BEAT’s MegaETH page, read on 15 November 2026, lists the points a user should know before bridging funds:

  • A single operator produces all blocks. If it stops or censors, users can force a transaction through Ethereum, but L2BEAT says that can take up to 12 hours.
  • Only one whitelisted address, which L2BEAT calls the Vanguard, can propose state roots. A faulty root can be flagged but not replaced by someone else, so the chain can halt until the Vanguard posts a correct one.
  • A Safe multisig with a 6-of-10 threshold, raised from 4-of-8 on 11 February 2026, can upgrade the bridge and the core contracts with no delay. Users have no exit window before an upgrade takes effect.
  • L2BEAT could not reproduce the hash of the Kailua proof program from published sources and marks this as critical; it also says the node source code is not available, so outside challengers cannot independently watch the chain.
  • L2BEAT warns that fewer than five external actors can submit challenges and writes that MegaETH is not even a Stage 0 project.

L2BEAT put the value secured by MegaETH at $85.40 million on 15 November 2026, down 5.30% in seven days. That count includes tokens minted natively on the chain ($56.12 million) and bridged from outside ($25.87 million and $3.39 million through the canonical bridge), which is why it is far above DefiLlama’s TVL, a measure of deposits in DeFi apps.

L2BEAT’s monitor also recorded two pauses in transaction data submissions on 29 October 2026, of about 15 minutes and 44 minutes, between 20:29 and 21:29 UTC, against a usual interval of about three and a half minutes.

MEGA says it wants to change much of this. According to CoinDesk and Crypto Briefing, it aims to complete its decentralisation changes by the end of March 2027, led by faster finality and sequencer rotation. The token page sketches rotation: operators would stake MEGA, be chosen by stake and performance, accept delegation and be slashed for faults, and the sequencer would move between regions to stay close to active users. No minimum stake, slashing rate or launch date has been published.

The MEGA token: 10 billion supply, 53% kept for KPI rewards

MEGA has a fixed supply of 10 billion tokens. CoinGecko’s MEGA page showed about 1.13 billion in circulation on 15 November 2026. The allocation below is from the official token page. The Block’s April report gave slightly different splits (53.3% for KPI rewards, 14.7% for venture investors, 9.5% for the team, 7.5% for the reserve, 5% for the public sale), and the token page itself warns that its chart is rounded.

AllocationShare of supplyTerms published
KPI rewards53%Released to committed holders when network milestones are attested
Community15%Echo, Fluffle, Sonar sale and mainnet campaign; unsold or undelivered tokens go to the Foundation
Venture investors15%No vesting schedule on the token page
Team and advisers10%No vesting schedule on the token page
Foundation and ecosystem reserve7%Held by the Foundation

Some sale tranches have known unlocks. Echo investors received 20% on 30 April 2026 and claim the rest of their vesting on Echo’s platform. Fluffle NFT holders received half on 30 April, with the other half vesting over six months. Public-sale tokens reached bidding wallets on 30 April 2026, or are due on 30 April 2027 for buyers who accepted a lockup.

How the KPI-triggered launch worked

MegaETH set no launch date for its token in advance. On 6 February 2026 the Foundation said the token generation event would come seven days after any one of three goals was met: $500 million of USDm, its stablecoin, in 30-day time-weighted circulation; ten apps launched; or three apps each earning at least $50,000 in fees over 30 consecutive days, according to The Block. The apps goal was met on 23 April, when ten MegaMafia apps had each passed 100,000 transactions in 30 days. MEGA started trading on 30 April 2026; Binance, KuCoin and Bitget opened spot markets at 11:00 UTC.

The same logic governs the 53% pool. Holders commit MEGA through Flux, and a pre-defined tranche is released when a milestone in one of four areas (reliability, performance, adoption, decentralisation) is reached and attested. MegaETH’s launch post put it this way: “The longer you commit, the higher your maturity and the greater your share of any rewards distributed.” For now the Foundation defines the milestones and attests them itself; token-holder votes and then oracle-based checks are planned for later phases. The token page gives no numeric thresholds for the remaining tranches.

The public sale: $1.39 billion bid for about $50 million of tokens

The public sale ran on Sonar in October 2025: a 72-hour English auction for 5% of supply, with a floor at a $1 million fully diluted valuation and a ceiling at $999 million. More than 50,000 bidders committed $1.39 billion, according to The Block, 27.8 times what was on offer, and the price was capped at $0.0999 per token. That held the raise to about $50 million. Kong told The Block in November 2026 that the sale brought in about $40 million, the gap reflecting discounts for buyers who accepted a one-year lock. Reports of a $450 million token sale in 2025 mix it up with the $500 million stablecoin pre-deposit of November 2025, which sold no MEGA.

What MEGA holders actually get

MEGA carries no share of network fees, and governance rights are planned, not live. Holders can commit tokens to Flux for KPI rewards. The token page says all rewards the Foundation receives from USDm are used to buy back and accumulate MEGA, “executed regularly”, but no amounts have been published. Staking for sequencer operators and Proximity Markets, server seats near the sequencer sold by auction, by locking MEGA or both, appear on the token page as planned products with no date. Crypto Briefing reported on 11 November that MEGA will consider changes to its tokenomics and explore programmatic buybacks, possibilities rather than commitments.

CoinDesk wrote on 14 November that the token had fallen more than 70% this year, and The Block put it more than 76% below its all-time high on 10 November.

USDm: the stablecoin meant to pay for the sequencer

MegaETH introduced USDm on 8 September 2025 as the chain’s own stablecoin. It is issued through Ethena’s stablecoin stack, the first version on the rails of Ethena’s USDtb, with reserves mainly in BlackRock’s tokenised Treasury fund BUIDL, held through Securitize, plus liquid stablecoins for redemptions. The plan was circular by design: reserve yield would cover the sequencer’s costs, so MegaETH could run it at cost and keep fees low. Kong said at the launch that “USDm means lower fees for users and a more expressive design space for applications.”

The supply figures show how far that plan is from working. DefiLlama’s stablecoin tracker counted about $165.8 million of USDm on MegaETH on 29 April 2026, a peak of $562.1 million on 3 May, $260.5 million on 19 May and $29.3 million on 18 July. On 15 November 2026 it counted $20.35 million. The $500 million launch goal was never reached, and the yield on $20 million covers little. MegaETH has not explained the fall.

Much of the spring liquidity went through Aave, which deployed a V3 market on MegaETH after a governance vote. The Defiant reported Aave deposits on the chain above $575 million in the days after the token launch. USDm was listed there with a 0% loan-to-value: it could be borrowed but not posted as collateral, and Aave’s risk adviser LlamaRisk expected the main use to be a loop of USDe collateral against USDm debt. On 15 November 2026 DefiLlama counted $4.70 million in Aave V3 on MegaETH.

MegaETH TVL and activity: what DefiLlama shows

DefiLlama tracks MegaETH as a chain rather than as one protocol. Its TVL series starts on 30 December 2025, climbs with the token launch to $245.59 million on 12 May 2026 and falls to $8.81 million on 15 November 2026, a drop of about 96%. Fees and trading volume read the same day point the same way.

MegaETH, DefiLlama (read 15 Nov 2026)24 hours7 days30 daysAll time
Chain fees$2,608$18,286$133,249$8.16M
Chain revenue$509$2,946$61,361$4.21M
DEX volume$795K$4.51M$18.7M$857M

What remains is concentrated in a handful of apps. Aave V3 holds $4.70 million on MegaETH, the Kumbaya exchange $1.98 million, Morpho Blue $0.92 million, the Feather vaults $0.52 million and Euphoria, a one-tap price-prediction app, about $0.19 million. RockSolid, the largest vault built around USDm, holds its $17.0 million on Ethereum rather than on MegaETH, and the canonical bridge counted $3.4 million on the Ethereum side.

MegaMafia shut, the rename to MEGA and a 20% staff cut

Much of the early activity came from MegaMafia, an accelerator that brought in about 20 teams over two cohorts. Those teams raised roughly $80 million between them, according to crypto.news. On 16 July 2026 Kong closed the programme, writing that it “was built on assumptions that no longer hold” and that “very little of that value has trickled to Mega”. MegaETH had taken no equity, governance rights or ownership in the teams. Several moved on: GTE built its own chain, Noise chose Base, HelloTrade moved toward Monad, and Cap went multichain after launching on MegaETH. The company said it would build first-party consumer apps instead.

The second turn came on 10 November 2026. The Foundation renamed itself MEGA and committed $30 million from its existing treasury to products running on MegaETH, starting with a market for AI inference in December. MEGA calls the approach “Purified Intelligence”: inference from different providers made interchangeable, so more suppliers compete on price. Providers and models will be named at launch. “There are still huge gaps between what providers charge for comparable inference,” co-founder Yilong Li told The Block. Kong tied the plan to the token: “This is our mandate, so wherever there are ways to utilize the MEGA token, we will.”

On Friday 13 November Kong announced the 20% cut. The people leaving worked on the blockchain, the wallet and the ecosystem; crypto.news adds communications. The founders, early employees and core engineers stay, and no headcount has been given. “Over the past few months, we have restructured MEGA to move in a new direction,” Kong wrote, according to crypto.news. According to CoinDesk, Kong wrote that staying a chain-only company would have meant ignoring the largest opportunity ahead. Outlets differ on how directly the cuts relate to the AI plan, and MEGA has not drawn the link itself.

An office tower, a robot, a downward arrow and a small group of people
MEGA pledged $30 million to an AI inference market on 10 November and cut about 20% of its staff three days later.

Funding and team

MegaLabs closed a $20 million seed round in May 2024 and announced it on 27 June 2024. Dragonfly led, with Figment Capital, Folius Ventures, Robot Ventures, Big Brain Holdings, Tangent and Credibly Neutral. Angels included Vitalik Buterin, Joseph Lubin, EigenLayer founder Sreeram Kannan and Cobie. The round mixed equity and token warrants at a fully diluted token valuation Kong put at no less than $100 million, The Block reported.

On 13 December 2024 MegaETH raised $10 million more on Echo, the investment platform founded by Cobie, in two tranches that filled in under three minutes; about 3,200 people from 94 countries took part. A Fluffle NFT sale followed in February 2025 and the Sonar auction in October 2025. The Block counts more than $80 million raised across these rounds, and reports that MegaETH bought back a 4.75% stake, equity and token warrants, from pre-seed investors in the summer of 2025.

The co-founders are Yilong Li, chief executive of MegaLabs and a Stanford computer science PhD; Shuyao Kong, chief business officer; and Lei Yang, chief technology officer. The legal home of the MEGA foundation has not been made public.

Audits and incidents

BlockSec reviewed MegaEVM, SALT and the stateless validator between 11 October and 28 November 2025. Its audit report page lists four high-risk and one medium-risk issue, one recommendation and nine notes. The medium issue and one high-risk issue, an EVM crash from a wrong gas cost in a KZG precompile, are marked fixed. The three other high-risk issues, all denial-of-service vectors, are marked “Confirmed”, which does not say whether they were resolved.

Zellic assessed the pre-deposit contracts, a USDC vault on Ethereum and a USDm distributor, in a report dated 20 November 2025 that lists one medium and one low finding. MegaETH has published no audit of its sequencer, and the Kailua proof system is RISC Zero’s software.

The best-known incident came before mainnet. On 25 November 2025 MegaETH opened its stablecoin pre-deposit with a $250 million cap and KYC through Sonar, Forklog reported. Sonar’s servers went down within a minute, and deposits failed because an identifier in the contract did not match Sonar’s settings; the fix took 23 minutes because it needed multisig signatures. The cap then filled in 156 seconds. The team had a transaction raising it to $1 billion fully signed in its Safe, and an outside user executed it about half an hour before the planned reopening, which Safe allows once the threshold is met. The cap ended at $500 million. MegaETH called the episode “unacceptable”, said the contracts worked as designed and funds were safe, and offered withdrawals, which fewer than 5% of users took.

MegaETH vs other layer 2s and fast chains

MegaETH sells latency, but liquidity has stayed elsewhere. DefiLlama’s chain TVL on 15 November 2026:

ChainTVL (15 Nov 2026)Note
Base$5.17BOP Stack layer 2 run by Coinbase
Hyperliquid L1$2.23BOwn chain for an order-book exchange
Arbitrum$1.31BOptimistic rollup, second to Base by TVL
Monad$0.90BHigh-throughput EVM layer 1
OP Mainnet$0.38BOptimism’s own chain
MegaETH$0.009BReal-time OP Stack layer 2

The closest comparison is Hyperliquid, which also built on fast order books and a small set of operators and holds about 250 times MegaETH’s TVL. Against Arbitrum, MegaETH is much faster but more centralised: Arbitrum lets anyone submit fault proofs and makes most upgrades wait behind a governance delay.

The main risks of using MegaETH

  • One sequencer and one proposer: if either fails or misbehaves, users wait for force inclusion through Ethereum (up to 12 hours) or for a correct state root.
  • Instant upgrades by a 6-of-10 multisig with no exit window, so a bad upgrade could reach bridged funds before anyone can leave.
  • A proof program whose hash L2BEAT could not reproduce, and node code that is not public.
  • Data on EigenDA rather than Ethereum, which adds a dependency on another network.
  • Thin activity: $8.81 million of TVL and $20.35 million of USDm on 15 November 2026, after most liquidity left.
  • A company in transition, with a new AI business, a staff cut of about 20% and no published vesting for team and investor tokens.
  • Three high-risk audit findings without a published fix status, and no public audit of the sequencer.

What to watch next

MEGA plans to open its AI inference market in December 2026, the first test of whether the $30 million pledge brings users to the chain. The decentralisation target, with faster finality and sequencer rotation, falls at the end of March 2027. On 30 April 2027 public-sale tokens held under lockup are due to reach their buyers.

Strengths and points of vigilance

Strengths

  • Mini-blocks about every 10 ms give apps a latency close to a centralised server.
  • Users can force a transaction through Ethereum if the sequencer censors or stops.
  • BlockSec reviewed the execution code and Zellic the pre-deposit contracts, with both reports public.
  • Backed by Dragonfly and others, with a $30M treasury pledge for products on the chain.

Points of vigilance

  • One sequencer and one proposer; L2BEAT says it is not even Stage 0.
  • A 6-of-10 multisig can upgrade core contracts instantly, with no exit window.
  • TVL fell from $245.6M (12 May) to $8.81M (15 November 2026), and USDm from $562M to about $20M.
  • No published vesting for team and VC tokens, and a company refocusing on AI after a 20% staff cut.

Team

Funding and audits

RoundAmountDateInvestors
Seed$20MDragonfly (lead), Figment Capital, Folius Ventures, Robot Ventures, Big Brain Holdings, Tangent, Credibly Neutral; angels incl. Vitalik Buterin, Joseph Lubin, Sreeram Kannan, Cobie
Community round (Echo)$10MAbout 3,200 investors on Echo
Public token sale (Sonar)About $50M cap ($1.39B committed); about $40M after lockup discounts, per the teamMore than 50,000 bidders

Roadmap

  1. 2024-06-27Round seed da 20 milioni di dollari per MegaLabs, guidato da DragonflyDone
  2. 2025-09-08Annunciata la stablecoin USDm con EthenaDone
  3. 2025-10-30Chiude la vendita pubblica su Sonar: 1,39 miliardi di dollari di offerte, circa 50 milioni raccoltiDone
  4. 2025-11-25Pre-deposito in USDC per USDm, tetto alzato a 500 milioni dopo un'esecuzione anticipata del multisigDone
  5. 2026-02-09Mainnet pubblicaDone
  6. 2026-04-30Lancio del token MEGA, sette giorni dopo il raggiungimento dell'obiettivo delle dieci app il 23 aprileDone
  7. 2026-07-16Chiude l'acceleratore MegaMafiaDone
  8. 2026-11-10La Foundation diventa MEGA, 30 milioni di dollari impegnati in nuovi prodottiDone
  9. 2026-11-13Tagliato circa il 20% del personaleDone
  10. 2026-12-01Mercato di inferenza AI previsto per dicembre 2026
  11. 2027-03-31Obiettivo per gli aggiornamenti di decentralizzazione: finalità più rapida e rotazione del sequencer
  12. 2027-04-30Consegna dei token della vendita pubblica con blocco

Crypto-assets are volatile and you can lose all the money you invest. Nothing published on Coindaro is investment advice.

Related