The term sheet language is clean. Almost deliberately so. "No pre-mine." "Four-year linear vesting." "Community-first distribution." I have now read three versions of this document in eighteen months, and they share the same six phrases — the ones that tested well with retail allocators who learned the word "unlock" sometime after the Terra collapse and never forgot the lesson.
Here is what the document does not say, at the same font size.
Thirty percent of the one-billion-token supply — 300 million units — is allocated to the team and its early backers. Vesting is linear. There is no cliff. The clock starts at the genesis block, not twelve months later. That works out to roughly 205,000 tokens entering the circulating float every day for 1,460 consecutive days, priced against a fully diluted valuation of $1 billion set by a $30 million Series A.
The whale didn't flinch at that arithmetic. Neither did the lead. That indifference — not the number itself — is the part worth examining.
Parallel execution is the most oversold engineering claim in the Layer 2 market right now, and the money is still arriving.
The pitch fits on a slide. The EVM is sequential by construction: every transaction mutates a single global state root, so block producers apply them one at a time, in order. Parallel EVMs break that serialization by predicting which transactions touch which storage slots, executing the non-overlapping ones simultaneously, and re-running the ones that collide.
The promise is an order of magnitude more throughput without asking developers to abandon the execution environment they already know. No Cairo. No Rust. Solidity, but faster.
That promise has attracted a specific class of capital — allocators who missed the first L2 cycle and are unwilling to miss the narrative that replaces it. Monad raised north of $200 million on the thesis. Sei re-platformed its entire architecture around it. Every quarter produces two or three new entrants claiming 5,000 to 10,000 TPS in a controlled benchmark.
This week's entrant is smaller: a $30 million Series A, two institutional leads I have seen on three other cap tables this year, and a $1 billion post-money. The public materials lean hard on the phrase "no pre-mine," which has quietly become the sector's most effective marketing token since "audited by three firms" lost its pricing power.
It also matters more in a sideways tape. When everything is bid, dilution is invisible — beta absorbs it and nobody reads the emission curve. In consolidation, the float is the only variable that reliably moves, because there is no trend to hide inside. Chop is where unlock schedules stop being a footnote and start being the price.
I have covered this market long enough to know the interesting data is never in the deck. It is in the emission schedule, the access-list coverage, and the identity of the accounts that will be liquid on day one.
What Parallel Execution Actually Buys You
EVM parallelism is not a hardware problem. It is a contention problem.
Block-STM, the model most parallel-EVM implementations borrow from, is an optimistic concurrency engine. Transactions are speculatively executed in parallel, each writing into a multi-version data structure. At validation, the engine checks whether any transaction read a slot that another transaction later wrote. If it did, the transaction is invalidated and re-executed. Throughput scales with the share of transactions touching disjoint state — and nothing else.
Here is what the benchmarks omit. On a live chain, that share is small. When I pulled storage-access distributions from three production L2s for a piece last year, 60 to 80 percent of gas was concentrated in a handful of contracts: the dominant DEX router, the largest lending pool, and the two or three stablecoin contracts that every routing path touches. Those are precisely the hot accounts that force re-execution.
A synthetic benchmark of 5,000 TPS, built from ERC-20 transfers across freshly generated wallets, therefore tells you almost nothing about mainnet behavior. Those transfers touch disjoint slots by construction. Run the same engine against a realistic DeFi workload and the parallelism ratio collapses toward 1.2x to 2x. Amdahl's law is not negotiable, and it does not care about your Series A.
That does not make parallel EVMs worthless. It makes them incremental. Incremental is a different word than the one on the slide.
The Testnet Number Is a Marketing Artifact
5,000 TPS. Every project has one. None of them are comparable.
A throughput figure is meaningless without three pieces of metadata: the workload, the access-list coverage, and the hardware. If the scheduler cannot see which storage slots a transaction will touch, it can only speculate, and speculation costs re-execution. That is why access lists under EIP-2930 and their successors matter more than any execution engine. A chain with 40 percent coverage and a clever scheduler will lose to a chain with 90 percent coverage and a mediocre one.
I have seen exactly one team publish its contention curve — the graph of parallel speedup against storage-overlap ratio. It showed a clean 8x at zero overlap and a flat 1.3x once overlap crossed 30 percent. They published it because it flattered the zero-overlap region. The 1.3x is the number that will describe their mainnet.
Watch for that metric when this new entrant ships. If the documentation leads with TPS and buries contention, you already know which side of 30 percent they expect to land on.
The Dilution Math Is the Actual Product
Back to the 300 million tokens.
A four-year linear unlock with no cliff is not a community-friendly structure. It is the opposite — a structure that maximizes insider liquidity in the earliest window, when price discovery is thinnest and exit liquidity is most expensive to source.
Consider the sequencing. On day one, the entire insider allocation is already vesting. The community allocation — whatever share of the remaining 700 million the team labels "ecosystem," "incentives," and "treasury" — is typically emitted over a longer horizon, often under its own vesting curve. For the first twelve to eighteen months, the marginal seller is disproportionately an insider, and the marginal buyer is a retail allocator attracted by the TPS number.
At 205,000 tokens a day and a $1.00 implied FDV price, that is roughly $205,000 of structural supply hitting the tape daily — before a single token of community emission. If the token trades at $3, the same schedule is $615,000 a day. Price does not change the token count. It changes the dollar pressure, and it changes it in the direction that hurts.
The chart lies; the ledger does not blink.
"No Pre-Mine" Is a Label, Not a Guarantee
Here is where the language does real work.
"No pre-mine" means the genesis block did not credit insiders directly. It says nothing about the foundation allocation, the treasury, or the "strategic reserve" — three line items that appear on nearly every cap table in this cohort and become functionally indistinguishable from a pre-mine the moment the foundation is a counterparty to its own grants program.
I have audited allocation tables where the foundation address, the treasury address, and the team vesting contract were all controlled by the same three multisig signers. Different labels. Same keys. That is not a pre-mine in the technical sense. It is a pre-mine in every sense that matters to a holder.
Governance is a silent coup, not a vote. With 30 percent of supply vesting from genesis and a community allocation still being emitted, the insider bloc does not need to seize anything. It simply votes with tokens the retail base will not hold for another four years. By the time the "community" controls enough supply to contest a proposal, the proposals that mattered have already passed.
The consensus view is that parallel EVMs are competing on throughput.
They are not. They are competing on distribution, and throughput is the vocabulary the sector uses to make that competition sound technical.
Alpha is not given; it is seized in the noise.
Look at where the leverage actually sits. OP Stack and ZK Stack are not differentiated by their proving systems in any way a deploying team can meaningfully evaluate. They are differentiated by which ecosystem can put a warm introduction in front of a chain's BD lead, which grants program wires faster, and which foundation answers a legal question inside a week. The winning stack is not the faster one. It is the one that signs more chains, and it always will be.
Parallel execution is heading for the same terminal state. The engineering becomes table stakes within eighteen months — there are only so many ways to build a conflict-detecting scheduler, and Block-STM's design is public. What will not commoditize is the sequencer's revenue share, the airdrop eligibility criteria, and the question of which three market makers agreed to provide depth at listing.
That is the information the term sheet is engineered to bury beneath the TPS number. The founders know it. The leads know it. The only party still pricing the throughput is the retail bid.
The contention curve is the tell. If it does not ship alongside mainnet, it does not exist.
Pull the foundation multisig and compare the signer set against the team vesting contract. Any overlap greater than zero means the "no pre-mine" label has already done its job.
Track the day-one liquid float — not the category, the address list. Everything else in the deck is a forecast. That one is an inventory.
Volatility is the tax on the unprepared. The schedule is published. Read it before the market does.
