Anatomy of a Winner

What $AI and $MarsCoin have in common — and the checklist it produces · Aug 29, 2026

Read this first. Both contracts are now verified — source read and live state confirmed via BSC and Robinhood Chain RPC. Neither is bespoke: MarsCoin clones Flap's FlapTaxTokenV3, $AI clones Whetstone's DopplerERC20V1. What remains unverified is what happens outside the tokens — both vaults are external addresses whose behaviour no contract enforces.

$AI Artificial Inu

“Long compute.”
Robinhood · paired to NVDA (NVIDIA)
$100.5Mmarket cap
23,075holders
992Msupply
2026-07-14first pool

Pools

PairLiquidity Vol 24hDEX
AI / NVDA$3.4M$5.4Muniswap
AI / WETH$1.0M$12.8Muniswap

Trajectory

DateHolders LiquidityMarket cap
Aug 239,783$1.4M$20.2M
Aug 2816,672$2.4M$55.7M
Aug 2920,767$3.4M$101.8M
Community Vault: 668.47 NVDA (~$146,295)  ·  6,394.51 NVDA (~$1,399,438)

Verified on-chain — $AI

Source read and state confirmed via Robinhood Chain RPC (chain 4663). $AI is a clone of Whetstone Research's DopplerERC20V1 launch template. The single most important fact: there is no tax, no fee logic and no vault anywhere in the token contract. Everything the site describes about the Vault happens outside it.

ClaimVerdictEvidence
No team allocation or vesting overhang The template supports pre-mints of up to 80% of supply. $AI uses none: vestedTotalAmount = 0, vestingScheduleCount = 0, and the contract holds 87 tokens. No insider unlock cliff exists. Genuinely clean.
Supply is being burned totalSupply is 991,721,653 of 1B — 8,278,347 retired (0.83%). Real, but small.
Anti-whale limit? isBalanceLimitActive = false and maxBalanceLimit = 0 — the feature was never configured, not merely expired.
Vault / burn enforced in code The token has only a permissionless burn() that burns the caller's own balance. “Burned outright or locked away permanently” is a policy someone executes, not a rule the contract enforces. The Vault is an external address with no on-chain guarantee.
Logic can't be swapped 44-byte Solady minimal clone — deployed bytecode 3d3d3d3d363d3d37363d73… hardcodes implementation 0x3be8b97f… (13,927 bytes, DopplerERC20V1). Same immutability as MarsCoin: the code cannot change, only what that fixed code permits the owner to do.
Ownership renounced owner() = 0xeb7c0347…, a 5,695-byte contract (itself owned by 0x21e2ce70…). Live, not renounced — the opposite of MarsCoin.
The owner can halt selling in two transactions. _beforeTokenTransfer reverts whenever to == pool && isPoolLocked. Right now pool() points at the dead address and isPoolLocked is true, so the only thing blocked is burning-by-sending — trading is unaffected and this looks benign.

But the owner can call unlockPool() and then lockPool(<the real LP>), after which every transfer into that pool reverts — which is every sell. There is no timelock on this and no renouncement. Nothing suggests intent to use it; the point is that the capability exists and is undisclosed.

$MarsCoin MarsCoin

“The first BNB coin paired to tokenized stock.”
BNB · paired to SPCXB (SpaceX)
$45.3Mmarket cap
holders
1000Msupply
2026-07-27first pool

Pools

PairLiquidity Vol 24hDEX
MarsCoin / SPCXB$989.5k$2.1Mpancakeswap
MarsCoin / USDT$350.9k$2.0Mpancakeswap
MarsCoin / USDT$317.1k$3.7Mpancakeswap

Trajectory

DateHolders LiquidityMarket cap
Aug 28$926.9k$38.4M
Aug 29$1.0M$50.1M
SPCXB Rewards Vault: 338.55 SPCXB pending — verified on-chain  ·  21.6K SPCXB distributed all-time (~$3.04M, self-reported)

Verified on-chain — MarsCoin

Contract source read and every claim checked against live state via BSC RPC. MarsCoin is not a bespoke contract — it is a minimal clone of Flap's generic FlapTaxTokenV3 template, so everything here applies to any Flap tax token.

ClaimVerdictEvidence
3% buy / 3% sell tax buyTaxRate() and sellTaxRate() both return 300 bps.
338.55 SPCXB in the vault balanceOf on the dividend contract 0xbba9d212… returns exactly 338.55 SPCXB. The marketing number is real and checkable.
Tax rate can't be raised later No setter exists in the contract, and owner() is 0x000…000 — ownership renounced.
Logic can't be swapped 45-byte EIP-1167 minimal clone; the implementation address is hardcoded in bytecode. No beacon, no proxy admin, not upgradeable.
Rewards run indefinitely taxExpirationTime is year 2126. The template supports an auto-expiring tax that flips rates to zero — this token effectively disabled that.
All trading funds the vault This is the finding. See below.
Only 27% of MarsCoin's volume actually pays the tax. The template runs a one-way state machine: BondingCurve → Migrating → TaxEnforcedAntiFarmer → TaxEnforced → TaxFree. During the anti-farmer window every pool is taxed. That window expired 2026-08-26 and the token is now in TaxEnforced (state 3), where tax applies only to mainPool — the MarsCoin/SPCXB pair.

Everything else trades tax-free, and that is now most of the activity: $2.10M taxed of $7.89M total. The three USDT pools carry $5.78M of volume and contribute nothing to SPCXB rewards. This is not disclosed on the site, it took effect three days ago, and it cannot be reversed — the state machine only moves forward and ownership is renounced.

Reusable screen: for any Flap tax token, read state(), mainPool() and antiFarmerExpirationTime(), then compute what share of DEX volume actually routes through mainPool. That ratio is the real yield, and it silently collapses the moment the anti-farmer window closes.

The mechanics are opposites

Both accumulate real equity from trading. What they do with it could not be more different — and that difference is the single most important thing on this page.

 $AI — accrue & burn$MarsCoin — accrue & distribute
Fee sourceLP swap fees on AI/NVDA and AI/WETH3% buy + 3% sell transfer tax
Stock side goes toCommunity Vault — retained, not claimableRewards vault → paid out to holders of 10K+
Meme side goes to“burned outright or locked away permanently”Not disclosed
Holder getsScarcity + a treasury narrativeSPCXB — actual tokenized SpaceX
Proven to date≈8.3M of 1B supply retired; Vault holds 668 NVDA21.6K SPCXB distributed (~$3.04M)
Failure modeValue is a story about a treasury nobody can touchYield decays with volume; tax deters traders

The DNA panel

Traits shared by both, turned into something we can run against smaller tokens before they win. ✓ present · ✕ absent · ? unmeasurable.

Trait$AI$MarsWhy it matters
Semantic fit with the stockThe meme is the company's thesis, not a random pairing. NVIDIA→compute/AI; SpaceX→Mars. This is what makes a ticker the Schelling point for its lane.
Owns its lane outright$AI leads NVDA by ~18×; MarsCoin leads SPCXB and is the only BNB name with a stock-native identity. Every confirmed winner we track consolidated its lane rather than splitting it.
Sustained holder growth, not a one-day spike?$AI: 9,783 → 20,767 holders in six days while price rose — demand broadened as it went up. BNB exposes no reliable holder count, so MarsCoin cannot be verified on this axis.
A vault that accrues the tokenized stockBoth accumulate real equity exposure from trading. This is the mechanic that does not exist anywhere else in crypto — fees denominated in shares.
Holders can actually claim itThe sharpest difference. MarsCoin wires a dividend contract into the token itself and pays SPCXB to holders of 10K+. $AI states plainly that “holders cannot redeem assets from the Vault” — and its contract contains no distribution logic at all.
Contract can't be changed against youInverts the naive read. Both are immutable minimal clones — neither's code can be swapped. The difference is what that fixed code permits: MarsCoin renounced ownership and has no tax setter; $AI keeps a live owner that can freeze selling in two calls. The smaller token has the stronger guarantees.
Real project surface (site + X + named identity)Both run a domain and an active X account and describe their own mechanics publicly. Most lane front-runners we scan have neither.
Standard ~1B supply, no supply games$AI 991.7M of 1B (≈8.3M retired); MarsCoin ~999.7M. Neither uses rebasing, reflections into supply, or mirror-to-share-count tricks.
Multi-pool depth, not a single pool$AI runs NVDA + WETH; MarsCoin runs SPCXB + three USDT pools. Single-pool tokens die when that one pool is pulled — as $Seyclops did, $11.9M → $0 in a day.
Graduated beyond the stock pair$AI's WETH pool now does 2.4× the volume of its NVDA pool, and other tokens ($SIT, $CLANKER) quote themselves against $AI. It became an asset in its own right. MarsCoin is still stock-anchored — arguably purer, but a lower ceiling.

Disqualifiers

Every one of these is something we watched happen this month.

Age under ~7 days$AU cleared every metric bar at 1 hour old; $Seyclops hit a $102M cap at 2.4 hours and went to $0 liquidity the next day. Both winners here are 30+ days and grew into it.
Volume ≫ market capMOONCOIN traded 23× its own cap in a day on $19k of liquidity. Bot loop, not demand.
Few, large trades$Seyclops did $19.4M across 1,011 trades — $19,189 each. $AI averages $362 and $BONER $104. Crowds move in small tickets.
Liquidity parked, not traded8 BNB tokens hold $8.7M on under $10k of daily volume between them.
Ticker squatting in the laneSix separate $HOTDOG contracts split the Costco lane and none of them won.

What this actually tells us

The pairing has to mean something. Both winners chose a stock whose thesis is the meme — compute for NVIDIA, Mars for SpaceX. That is what makes a ticker the obvious answer for its lane, and lane consolidation is what every winner we track has in common.

A vault of tokenized stock is the genuinely new primitive. Nowhere else in crypto do trading fees accrue in equity. Both of these built one. Whether holders can claim it is the open question: MarsCoin pays out, $AI explicitly does not — and $AI is the larger of the two, so the market has not yet priced distribution as superior.

Graduating beyond the stock pair may be the real ceiling test. $AI's WETH pool now trades 2.4× its NVDA pool, and other tokens quote themselves against $AI. It stopped being a stock-paired meme and became an asset. That is the step from $100M toward $1B, and MarsCoin has not taken it.

Neither is a business. No revenue, no product, no cash flow. $AI says so outright. These are attention assets with an equity-accrual mechanic bolted on — which is interesting and genuinely novel, but should not be mistaken for a protocol earning money.

Caveats and gaps

P2 Ventures · sources: project sites (artificialinu.com, marscoinbnb.com), DexScreener, Robinhood Chain Blockscout, and our own daily snapshots · Master board