How Exit Valley works

You buy one NFT: a fictional AI company, on one of two sides. It does nothing at all until you stake it, and then it emits $EXIT every second it stays staked. You spend that $EXIT on GPU upgrades, which permanently multiply what that one NFT earns — and the same spend is what scores points for your company’s faction.

Everything is on chain and nothing is custodial. Your NFT is a standard transferable ERC-721 while it is unstaked, the yield is a real ERC-20, and both are yours to sell whenever you want.

There are exactly four things the game asks you to decide: which company to buy, when to stake it, how far up the GPU ladder to spend, and when to take your yield out. The rest of this page is what each of those does.

The game is not live the moment it is deployed. Two dates are announced — when the mint opens and when emissions begin — and the second must fall after the first. The mint then opens by itself when its date passes, with no transaction from anyone, and keeps running through the whole season. Arming emissions is the one step that still needs a signature, and it needs a multisig rather than the day-to-day key.

Either date can be moved until it arrives, and is frozen from that instant. For emissions the freeze lands when they are armed rather than when the date passes, so a late operator cannot brick the game — and when they finally arm it the start is clamped forward only, to whichever is later of the announced date and the moment of the call. You can be made to wait, but you can never be started early on a date you did not see, and lateness is emitted on chain.

Supply is hard-capped at 5,000 NFTs — 500 per company across 10 companies — and the mint closes itself the moment the cap is reached. One company selling out leaves the other nine open. The mint also refuses once emissions are over, so the curve can never charge full price for an NFT that provably cannot earn.

Each company has its own price curve and they are independent. The first mint of a company costs 0.0001 ETH and every mint after it adds a fixed increment, so the last one costs 0.0005 ETH. The curve is a pure function of how many of that company have been minted: it does not decay, it does not reset, and waiting does not make it cheaper.

Once the mint has closed the only way in is buying from someone who already holds one. There is no second mint.

$EXIT is minted once, at deploy, and never again by the game. 10,000,000,000 tokens are created and split three ways. Yield is paid out of the yield reserve contract, by transfer — withdrawing does not mint anything, so no amount of play can inflate the supply. That contract has no owner, no withdraw function and no rescue: the only way a token leaves it is a farming payout.

There is no team allocation. The team is paid by the mint proceeds, by every $EXIT spent on a GPU, and by the fast-withdrawal fee — all of which only exist if people actually play. Nothing is set aside up front.

Nothing in this game burns anything, either, and no number on any screen should ever be labelled that way. $EXIT spent on a GPU is collected: all of it goes to the protocol treasury, and none of it returns to the yield reserve. The reserve therefore only ever goes down — it started at 9B and nothing in the game pays back into it, which is why the GPU ladder and the length of the game are the same question.

Holding an NFT earns nothing. You have to stake it. From the moment emissions begin the protocol pays out a fixed 80,000,000 $EXIT per day — not per NFT, but in total, split across everything staked in proportion to GPU weight. Your NFT earns its share of that budget, so what you make depends on what everybody else is doing: the same NFT earns less the more company it has. The yield accrues to the token, not to your address — which is the detail everything else in this section follows from.

A staked NFT transfers normally — but the transfer closes the stake and forfeits its un-withdrawn $EXIT. That is true of an old marketplace listing too: it can still be filled, at the price it was signed at, and the accrual is destroyed in the same transaction. Withdraw before you list. A marketplace still gets its blanket permission — that is deliberate, so one staked NFT does not stop you listing the rest of your wallet — but a single-token approval is refused while staked, and staking clears any that were already outstanding. The one transfer the contract refuses outright is on a token with a withdrawal request in flight.

There is no claim button. The only route from accrued yield to a spendable balance is: request a withdrawal, wait 5 minutes, then take it. A claim that parked yield in your account balance would make “claim, then unstake” a way to keep money you were about to forfeit, so the button does not exist.

A request names the tokens it settles and freezes what each one is worth at that instant, so the amount you are quoted is exact rather than a moving target. You can cancel it at any time, including after it has matured — cancelling puts every amount back onto the token that earned it.

If you do not want to wait, you can pay 35% to collapse the clock and be paid immediately: 65% reaches you and the rest goes to the protocol treasury. It does not stay in the reserve, and it does not fund anyone else’s yield. Skipping changes the timing and nothing else — the same amounts are settled, from the same tokens, with the same consequences. Once a request has matured the skip is refused outright, because withdrawing is then free.

Unstaking destroys yield you have not withdrawn. That is the whole design, and it is why nothing is escrowed: a pending withdrawal is a claim on the reserve, not a funded balance sitting in the farm. You cannot unstake while a request is open — cancel it first, which puts the money back on the token where unstaking can then destroy it. The destruction is always something you perform deliberately.

There are 6 rungs above stock hardware, from New Hire up to Founder at 10× base. Each one is a bigger multiplier on everything that one NFT earns.

The tier is stored per token, never per wallet. Upgrading one company does nothing for the others and you climb each ladder separately — and because it belongs to the token, it survives a sale. Buy an upgraded NFT on the secondary market and you keep the tier somebody else paid for.

Read the last column before you spend anything. It is the cost of a rung divided by the extra yield that rung adds, and it rises all the way up the ladder — every rung is a worse deal per unit of yield than the one below it. There is no bargain hidden near the top and no toll rung to get past. The cheapest payback available to you is always the next rung up, which makes the decision an honest one: how long do you expect to keep playing?

Rungs are sequential. You cannot buy Founder from stock hardware; the cumulative column is what reaching a tier actually costs.

There is no maintenance. A rung, once bought, is yours for good. Nothing decays, nothing lapses, no weekly fee keeps it alive, and there is no way for an NFT to be downgraded. Upgrade spend is not burned either — all of it is collected by the protocol treasury, as described above.

Upgrading closes once the game is ended. A tier bought after emissions have stopped would multiply nothing.

5 companies fight for the USA and 5 for China, and the side is a property of the token, decided by which company it is. There is no per-wallet faction, no lock at mint, and nothing to sign up for.

So a wallet is not a faction. Yours can hold both sides at once, or neither, and nothing anywhere in the game asks you to pick one. Hold one of each and you get both war rooms, and you score for both sides.

A side’s war score is the $EXIT its NFTs have spent on GPU upgrades. Not what they hold, not what they have earned — what they have committed. Passive accrual scores nothing, which is the point: the war measures play.

Prize rounds are funded in ETH and are opened and finalized by the operator. There is no fixed cadence and no timer on chain; a round exists because somebody started it. Winners are published as a merkle root, and claiming a prize registers it rather than paying it out — the amount then unlocks linearly over 7 days.

The end date is on chain, and it is the point of the design. The protocol emits a fixed budget each day that halves every 45 days — 80M, 40M, 20M, then 10M $EXIT per day — and pays nothing at all after 180 days of emissions. How long the game runs is an input, chosen up front and enforced by constants no operator can move.

Upgrading cannot make the protocol emit faster. That is the whole difference from a per-NFT faucet, and it takes some getting used to: the ladder no longer changes the size of the daily pot, only how it is divided. Buying a rung takes share from the tokens that did not buy one — which also means that if everybody upgrades together, everybody ends up exactly where they started, poorer by the cost of the ladder. Upgrading is a race against other players, not a private multiplier.

So the reserve cannot be drained early by heavy play, and the two numbers are now independent: the reserve holds 9,000,000,000 $EXIT, of which the game is allotted a hard quota of 6,750,000,000 — exactly the schedule above, and not a token more. The rest is reserved for other consumers and the game cannot reach it.

This is a satire about overinvestment, and it is also real money on a real chain. These are the things the code does not guarantee, stated plainly rather than left for you to discover.

Pending yield is not funded. Nothing is escrowed for you. A pending withdrawal is a claim on the reserve, so if the reserve is empty when you go to take it, it does not pay. Withdrawing earlier is strictly safer than withdrawing later, and the further into the game you are, the more that is true.

Prizes depend on an operator. Faction rounds have no schedule, no guaranteed prize, and no guarantee that a round which opens will be finalized. The winning set for a round is published as a merkle root, and the proof you need in order to claim is produced off chain.

The game can be ended deliberately. Ending it is a single administrative call and it cannot be undone. It stops emissions and closes upgrading for everyone, permanently.

Emissions being announced is not emissions being on. Nothing on chain executes at a timestamp. The announced date is a commitment that emissions will not begin earlier; somebody still has to send the transaction that arms them, and until they do, every staked NFT accrues exactly zero. A date in the past with emissions still off is a real state, not a bug.

A listing made before you staked stays visible. If you list an NFT and then stake it, the listing does not disappear — a buyer’s attempt to fill it just fails on chain. Cancel the listing before you stake.