Fees, Economics and the Token

What a transaction costs, what a block costs to settle, the size of today's subsidy, the five things that close it, and what the token is and is not for.

What a fee is made of

A Layer 2 fee is not just execution. It carries a share of what it costs to publish data and settle on Solana, plus the cost of proving. Any fee model that ignores those is subsidised, and subsidies end.

Settlement cost is per batch, not per transaction. A thousand transactions in a batch each carry a thousandth of it. That is the entire economic argument for a rollup, and it also means per-transaction cost depends on how full batches are, which is why a single headline fee figure is misleading.

What a transaction costs today

A flat 5,000 lamports, paid in SOL, the same order as a Solana signature fee. It is a placeholder for the composition above rather than a market price.

SOL is the gas asset and stays the gas asset. There is no fee token, no fee discount in a token, and no token bid in any auction. A user never needs to acquire anything to use the chain, and the asset they pay in is the asset the operator's own costs are denominated in, so there is no oracle and no exposure between the two.

What a block costs

The devnet chain earns nothing by design and pays for every block it produces. Measured by the node itself:

Three things follow.

Rent, not fees, is the cost of running a rollup on Solana. It is roughly 164 times a block's fee revenue at this chain's rent schedule. It is sunk because nothing closes a root or batch record: the commitment chain is the chain.

A block needs about 820 transactions at the flat fee to pay for its own settlement. Below that the chain is subsidised, and it is subsidised today by about three orders of magnitude. That is fine for a test chain and it is disclosed rather than hidden; it is not a chain anyone should call self-supporting.

The floor is not the flat fee. The measured data-availability share alone is about 280 lamports per transaction on modelled sustained traffic, derived from the compression figures and Solana's posting mechanic. Sparse traffic compresses worse and costs more.

Two facts make the whole picture tractable. Blocks are produced only when there is work, so an idle chain costs nothing. And the dominant cost is rent, a function of bytes rather than congestion, so it is predictable and, unlike a fee, recoverable.

Five things that close the gap

1. The fee is kept, not burned

It is credited on the Layer 2 to the sequencer identity the genesis declares, and leaves through the bridge like any balance. A burn is what a Layer 1 does with its base fee because there is nobody to pay. A rollup's sequencer pays real Layer 1 costs, and every rollup that covers them keeps the fee. Burning while paying rent is the one combination that cannot stand.

Implemented, and activated from a per-chain block height so earlier blocks replay exactly as they were: the current chain burned up to block 100 and has kept the fee since, a switch it crossed on 10 September 2026, and each block reports which rule it used. The fee step changes shape with it, from one leaf to two, so the verifier gained a paid-fee class beside the burn class.

The sequencer's balance sits on the Layer 2 until swept through the bridge, which is an operational hot-key exposure. The sweep should be frequent and the destination a cold key on Solana.

2. The fee is composed rather than flat

A data share from the bytes published, a settlement share from the block's fill, and a margin. This is computed today and charged by nothing, and the arithmetic shows why it is not ready.

The settlement share is a division by the fill, so it is largest exactly when a block is emptiest. On the current block shape, a one-transaction block prices at 4,487,240 lamports and an 800-transaction block at 6,066. The payload compresses with fill too, 223 bytes per transaction at one and 89 at fifteen, which is the same effect from the other side.

Both the margin and the fee floor default to zero, which is the absence of a published price rather than a price, because this step requires disclosure before charging.

3. Rent is reclaimed

Finalized root records are closed after a retention period and their rent returned to the root's proposer; sealed batch records are closed after a longer one, with the bytes archived first. This turns the dominant cost from sunk into float. Both are built and gated by tests. Neither is deployed.

Retention is deliberately two bounds at once, a count of newer final roots and a floor in slots, because closing a root ends the ability to prove new withdrawals against it. Retention is therefore a proving deadline, and a count alone is a deadline that shrinks as block cadence rises.

4. Ordering is sold, in SOL

A priority-fee auction pays the sequencer for the one thing it has that costs it nothing to give. Planned. Priority-fee instructions from Solana are not honoured today.

5. Services

Keyed RPC and explorer APIs, indexing, and running chains for others with the deployment tooling that already exists. This is revenue that does not wait for transaction volume.

The canonical bridge charges nothing in either direction beyond Solana's own fee. The fast exit earns a spread, in the asset withdrawn, and the operator earns there as a provider or through a share of it.

When Solana's own fees spike

Base-layer fees are not constant. When they rise sharply an operator has three options, and the time to pick one is in advance rather than during the incident.

Solieum will publish which one it does, and the threshold at which behaviour changes, before mainnet. An undefined policy here becomes an improvised decision during exactly the wrong week.

The principles the fee model answers to

• Fees are charged in SOL. No token is required to transact.
• Fees track real cost. No figure that only holds while someone is subsidising it. Today's flat fee does not, and the measured gap is stated above rather than hidden.
• The cost model is published, so a quoted fee can be checked rather than trusted.
• Fee changes are announced before they take effect.
• No hidden spread between what is charged and what is paid to Solana.

The token

The token does not exist. It is not issued. There is no address, no supply, no allocation and no sale. Until an address is published on the official site, every address claiming to be a Solieum token is a scam.

What has been decided is what it would be for, because that decision constrains the protocol being built now.

The line, and why it is drawn there

Anything that is a fee or a collateral securing user funds is SOL. Anything that is a reward, incentive, membership stake or vote is the token.

Why bonds stay SOL. A bond's job is to be worth enough to pay the challenger on the day a fraud is found, which is the day a protocol token is least likely to be worth anything. A guarantee that weakens exactly when it is needed is not a guarantee.

Why gas stays SOL. Making the token the gas asset puts a bridge-then-swap step in front of a user's first transaction, creates an exposure between the fee asset and the Layer 1 cost asset, and hardens the regulatory posture, all for gas revenue that is negligible. The large rollups settling to Ethereum charge gas in ETH and are governed by their tokens. This follows that pattern.

What it would govern, and what it must never touch

Governable, because it is economic: fee margins, the bridge fee, fast-exit fee caps, treasury spending and grants, reward schedules and their vesting, retention windows within a floor the code enforces, and admission to a staked sequencer set.

Not governable, because it is the security of the chain and lives in code changed by redeploy: the challenge window and its unit, the air gap and who may deny a root, the dispute authority, the one-step classes and their boundaries, the state commitment, and the supply rule. A vote that could shorten the window or move the dispute authority would make the token the security, which is precisely what it must not be.

Program upgrade authority stays with the operator's keys until a separate decision hands it to a governance-controlled multisig with a timelock.

Three rules that keep the rewards honest

1. No fee-path role. Gas is SOL, priority bids are SOL, the bridge fee is SOL.
2. No collateral role for user funds. Bonds that secure funds or pay a challenger are SOL, and stay SOL even if the token is worth more.
3. Rewards are paid for settlement that held, never for settlement promised. A proposer's reward is pending until the window closes clean, and a refuted root pays the challenger instead, along with the bond.

When it would be issued

When at least one of these is true, and not before:

1. Sequencing opens to staked outsiders.
2. Verifier coverage needs paying rather than volunteering.
3. Fast-exit liquidity needs bootstrapping beyond what SOL fees attract.
4. Decisions on the economic surface need a constituency wider than the operator.

Until then the chain runs on SOL, the roles are the operator's, and this paper says no token exists. Non-transferable points for early verifiers, challengers and providers may be recorded before issuance and honoured at issuance; that is a decision for the issuance record, not a promise here.

Supply, emissions, allocation, vesting, the mint address, and the chain it is minted on are deliberately undecided. Each is a decision with reasons, recorded when taken.