$BEM 与 PoD 挖矿$BEM and PoD mining

What is $BEM

$BEM is the incentive token of the TapeOut ecosystem, deployed on BNB Chain as an ERC-20 with a 21,000,000 supply. No presale, no team reserve: only the mining contract can mint. LayerZero later added 1:1 locked minting between BNB Chain and X Layer.

Basics (measured on chain on 2026-09-25):

"Burning" does not shrink total supply: the BEM contract has no burn or burnFrom function, so totalSupply is the cumulative minted amount; what people call burning is sending BEM to the dead address 0x…dEaD, which does not reduce totalSupply.

Bridging BEM between chains

Since 4 October 2026 BEM moves between BNB Chain and X Layer: 1:1 lock-and-mint over LayerZero, settling in one to two minutes, with the contract source verified on both block explorers and bridging available on the official site.

Bridging maps the same asset between two chains — what is locked on the source chain is minted on the target — so it is not a second token. Mining emission still happens on BNB Chain only; X Layer receives bridged BEM but produces none.

PoD: the contest is design, not electricity

Bitcoin uses PoW (proof of work): everyone solves the same puzzle, first to finish wins. TapeOut uses PoD (proof of design): no GPUs, no mining hardware, no staking — the better circuit wins. Difficulty rises naturally through competition, and copying someone else's circuit yields diminishing returns.

Hashpower = the cost you burn forever, plus how much better your design is than the reference, times the processor factor. — tapeout.net/pod/

What counts as a miner

A community "miner" is a circuit NFT taking part in PoD mining. A qualifying miner meets three rules:

1. On a supported processor: only circuits on Behemoth (#1) and TapeOut (#30) can mine.
2. No references (REF): the official algorithm page states that in this version a circuit with references is fully disqualified and cannot even enter the unverified pool.
3. Under the gate-count gate: too many gates and the on-chain spot check cannot finish, so the circuit cannot claim a task and lands in the unverified pool; the exact cap is shown on the mining page.

Why ban references? Referencing someone else's circuit means you never burn extra transistors, yet the circuit area still counts the referenced part. Allowed, a shell that burns almost nothing could inherit another designer's score. The official audit measured on a local chain that 14 zero-burn shells cut a real designer's share from 100% to 28%. Hence the rule: every unit of hashpower must sit on an equal, irreversible burn of transistors.

The mining flow

The official task bank holds 267 tasks, each with a verified reference implementation on chain. Design a circuit to a task spec in the canvas, then walk the flow:

1. Design: draw the circuit to the task spec. If you cannot draw, each of the 267 tasks ships with an official template, so one-click tapeout can produce a miner.
2. (Optional) Commit the design: call commitDesign to register a commitment so nobody can front-run your transaction. It produces a backup containing a salt (a random number only you know); do not share it before mining.
3. Tape out: burn transistors and receive a circuit NFT.
4. arm (reserve an anchor): wallet signature #1.
5. start (mine): wallet signature #2, which must confirm within 64 blocks (about 29 seconds). The contract immediately draws a few official test vectors (input to expected output) and checks your circuit bit by bit; the count scales with circuit size but is never less than 3.
6. claim: withdraw anytime, no lock-up.

Two pools:

• Verified pool (99%): circuits that claimed a task and passed the check, paid by hashpower H.
• Unverified pool (1%): circuits that claimed no task or exceeded the gate-count gate, paid by b*×P (you still get the cost part without solving a task).

The hashpower formula

The formula looks scary but only asks three things: how many transistors this layer burns (b*, paid in BNB), how much better your design is than the reference (q), and which processor you are on (P). The official formula page defines it as follows (symbols copied, plain words added):

Where:

• K_task: the cost basis of this task reference implementation.
• C_ref: the cost of this task reference implementation, frozen by the contract after running vectors when the task is created; the author cannot fake it.
• P: the processor factor, Behemoth = 6, TapeOut = 1. The same design on Behemoth yields exactly 6x hashpower.

Best first and the design premium: the premium K_task·q goes only to the best first. Per task and per processor there is only one such circuit — the one with the smallest design cost C. Whoever builds a smaller C takes it over; the previous holder drops to non-best and keeps only the cost basis:

• best first: H = (b* + K_task·q) × P
• non-best: H = b* × P

If your design is the same size as the reference (same cost basis, q = 1), the best first gets exactly 2x the hashpower of a non-best. In general the multiple is (b* + K_task·q) ÷ b*. When two designs have identical C, the tie breaks by valid commit > earlier commit > earlier tapeout.

Challenge and revocation: anyone, at any time, can submit a counterexample vector proving a miner computed an official vector wrong, with no stake required. Once upheld, that miner is permanently revoked. A separate "kick" only removes the best-first status and design premium; base mining continues.

Emission and halving

From the official formula page:

• Hard cap: 21,000,000 BEM (8 decimals), minted only by the mining contract.
• Initial emission: 7,200 per day.
• Halving: every 210,000 × 600 seconds, about 3.995 years.
• Unmined time: that emission is forfeited forever, never backfilled or rolled over.
• No transfer tax: on launch day (Aug 21) the founder dropped the planned 0.2% BEM transfer tax and related ownership, keeping only a 30-minute deploy lock window.

Network mining opened on 2026-08-21 21:45:32 (Singapore time).

Break-even math

Four steps: work out your hashpower, your daily BEM, your total cost, then divide. The hard part is not the arithmetic — it is that every input changes daily.

1. Your hashpower H: use the formula above. A non-best circuit is just b* × P.
2. Your daily BEM: daily BEM ≈ 7,200 × 99% × H ÷ (network verified hashpower + H).
3. Your total cost: transistor spend (unit price × burned count) + tapeout protocol fee (0.0002 BNB per call on chain as of 2026-09-25) + gas for tapeout and mining + other fees.
4. Static break-even days: days ≈ total cost ÷ (daily BEM × BEM price).

These are static estimates: they assume network hashpower, the token price, and your best-first status all stay put. In reality all three move daily, so leave margin — double the hashpower and take a conservative price.

A worked example (sample numbers)

Everything below is sample data, not real numbers, only to show the steps.

Step 1: hashpower (the circuit burns 40 NANDs and 4 LATCHes, area A = 68, depth d = 5; the task has C_ref = 10,000 and K_task = 60, on TapeOut, P = 1)

• cost basis b* = 40 + 1×4 = 44
• cost C = 68 × 5³ = 8,500
• quality q = 10,000 ÷ 8,500 ≈ 1.176 (within 1/4 to 4, no clamping)
• best first: H = (44 + 60 × 1.176) × 1 ≈ 114.6; non-best: H = 44 × 1 = 44

Step 2: daily BEM (assume network verified hashpower = 2,000,000, so the verified pool pays 7,128 per day)

• best first: 7,128 × 114.6 ÷ (2,000,000 + 114.6) ≈ 0.408 BEM/day
• non-best: 7,128 × 44 ÷ (2,000,000 + 44) ≈ 0.157 BEM/day

Step 3: cost (assume transistors at 0.0002 BNB each, protocol fee and gas together 0.003 BNB)

• 44 × 0.0002 + 0.003 = 0.0118 BNB

Step 4: break-even days (assume BEM = 0.001 BNB)

• best first: 0.0118 ÷ (0.408 × 0.001) ≈ 29 days
• non-best: 0.0118 ÷ (0.157 × 0.001) ≈ 75 days

Sensitivity: double the network hashpower to 4,000,000 and the best-first break-even doubles to about 58 days; halve the token price and it doubles again. If your best-first status is taken over, use the 75-day row.

Plug in your own numbers

How to compare miners? Community author Benson suggests a handy idea: do not just compare the price of one miner — compare how much BNB buys 1 BEM/day of capacity.

Where $BEM flows: fees, treasury and burn

The official site states (checked 2026-09-25): protocol fees go into the treasury and are swapped into BEM, and 80% of the treasury BEM is burned periodically (sent to an unrecoverable dead address). Fee sources include container transfer fees, tapeout protocol fees and the official market 1% fee. Homepage promo slots are bid on in BEM, and the BEM spent goes into the treasury.

• Circuit containers: fees for creating a container, transferring a container out and transferring assets out go straight into the treasury, are all swapped into BEM, and 80% of that is burned periodically.
• TapeHub launchpad: tapeout revenue, the 1% of launch pool tokens projects commit, and the BNB from buying 3x fuel all go into the TapeHub treasury; 50% of treasury revenue is used to buy back BEM and burn it permanently; projects that make their pool in BEM get a bigger push.
• Official market: the community timeline says that since Sep 21, 50% of fees are used to buy back and burn BEM (not independently verified).

All this burning only sends BEM to the dead address and does not reduce totalSupply.

Where to trade

• On-chain: the BEM/USDT pool of PancakeSwap V3 (the largest DEX on BNB Chain), 0x2f5e…ea38.
• CEXes: the community says MEXC, LBank, Gate, URBit and Ourbit listed it from Sep 14; the MEXC listing announcement and LBank listing record back this up.

Before buying, verify the contract address 0x5ce0…695a. Other tokens with "TapeOut" in the name (such as TAPEOUTX) are not BEM and have nothing to do with the protocol. Aggregator sites can lag badly; trust on-chain and DEX data.

V2 preview (not live yet)

The founder's Aug 26, 2026 V2 preview described a future economic model; as of 2026-09-25 it is not live yet and details may change:

• Component slots: third-party developers can build new components by protocol rules (such as BNN binary neural networks or LUT lookup tables) and list them in the "device warehouse" after official review.
• Device warehouse: what a processor can build depends on which devices it has bought, and buying devices requires burning BEM, giving BEM another use.
• 1% protocol fee: when users mint on a V2 processor, the protocol takes 1% of the maker BNB revenue; when minting a third-party component, half of that 1% is swapped into BEM and paid to that component developer, with base transistors excluded.

The V2 BNN component already has a prototype: 34,048 binary weights stored in the contract, 256-bit inputs, 128 neurons, and 93.15% accuracy on handwritten digits (official number).

Sealing: mining rules are locked

Sealing is the admin permanently throwing away the key to change the rules. The mining contract was sealed in three on-chain-verifiable steps:

1. 2026-09-12 16:08: the admin submitted the seal to the timelock contract (TimelockController), requiring a 48-hour wait, visible to everyone in advance; 25 minutes later the founder posted explaining why he gave up control.
2. 2026-09-14 17:38: the seal executed on chain. The founder announced that "the PodMining contract now has isSealed = true and no owner; BEM belongs to everyone".
3. 2026-09-25 checked: PodMining isSealed() returns true and owner() is the zero address.

From then on the mining rules, emission curve and 21 million cap cannot be changed, not even by the team. Note: only the mining contract is sealed (the transistor market is sealed too); it does not mean the whole protocol is frozen.

Risk note

$BEM is volatile; mining requires spending BNB on components first, and returns depend on design skill and network competition. The above is public ecosystem information, not investment advice.