# Cost plan — a sovereign, quantum-hardened Layer 1 for $2,000 down and $2,000 a month

Prepared 2026-09-23 for feature #792. The design already exists as the EQ Chain specification (F0, 2026-08-19): a sovereign CometBFT + Cosmos SDK chain with the EVM module, SLH-DSA validator keys, ML-DSA accounts, five permissioned genesis validators, and phases F1 devnet, F2 public testnet, F3 audit + mainnet, F4 EventTrader rail. #792 adds ET10 as the settlement asset (a canonical bridge, since ET10 lives on Ethereum) and an explicit open-to-builders posture, which the EVM module already gives.

## The headline: $2,000 outlay, $2,000 a month, full implementation

**Recommended lean path — $2,000 one-time, $2,000 per month all-in.** Full implementation through F4: devnet, public testnet, permissioned mainnet with Cymetica-run validators, the ET10 bridge, and the EventTrader rail. Engineering runs on the agent seats the platform already pays for. Assurance is staged instead of bought up front.

**Full-price path — $200k–550k one-time, $15k–25k per year.** Same chain, with an external audit of the whole post-quantum boundary and a MiCA/CLARITY counsel package bought before mainnet. Those two items are 90% of the bill. The code is the cheap part.

## Where the money goes, phase by phase

- **Stage 1 — provenance rails** (attestation registry on Ethereum, API, page, licence references). Full-price $500–1,500. **Lean $300**: agent build inside existing seats; only mainnet gas is paid in cash.
- **F1 devnet** (Cosmos SDK + EVM fork, vendored liboqs, pqauth module, ML-DSA / SLH-DSA precompiles, SLH-DSA validator keys, KAT + differential + reproducible-build CI). Full-price $5k–20k. **Lean $0**: consensus-critical Go written by agent sessions under two-approval review, as the spec already requires.
- **F2 public testnet** (5 validators + sentries, 2 public RPC nodes, explorer, faucet, docs, SDK). Full-price $3k–10k plus $1–2k/mo. **Lean $400 plus $900/mo**: Machine B hosts one validator and the explorer; six modest VPS hosts across three providers cover the rest.
- **ET10 bridge** (lock on Ethereum, mint on chain, relayer). Full-price $3k–8k plus gas. **Lean $600**: mainnet deployment gas and relayer float; code from existing seats.
- **F3 assurance** (external audit of the PQ boundary, retests, published report). Full-price $150k–400k. **Lean $500/mo**: staged — in-house gates now, a rolling public bug bounty from launch, an external audit bought from chain revenue before outside validators or third-party value at scale.
- **F3 legal** (MiCA classification + CLARITY analysis with counsel). Full-price $30k–100k. **Lean $0 until token economics**: the F0 compliance gate already passed; counsel is engaged when the ET10-as-gas decision is packaged.
- **F4 rail** (deposits, withdrawals and CLOB settlement on the chain). Full-price $2k–6k. **Lean $0**: same pattern as the Robinhood Chain and Arc integrations, each one to two session-days.
- **Ongoing** (hosting, monitoring, agent maintenance, bounty pool). Full-price $15k–25k/yr. **Lean $2,000/mo**, broken down below.
- **Total to a running mainnet with the rail live: full-price $200k–550k, lean $2,000 outlay plus $2,000 a month.**

Agent compute is treated as sunk on the lean path because the platform already ships every feature through the same Claude Code seats. On the full-price path it is priced at $100–250 per heavy session-day across roughly 80–150 session-days.

## The $2,000 outlay

- **$600** — Ethereum mainnet gas: bridge contracts, attestation registry, contract verifications
- **$400** — Host provisioning, first-month deposits, TLS and DNS
- **$300** — Stage 1 provenance rails go-live
- **$700** — Contingency: re-deploys, an extra RPC node during the chaos gate

## The $2,000 a month

- **$900** — Six VPS hosts (4 vCPU, 8–16 GB RAM, 400 GB NVMe) plus egress
- **$100** — Monitoring, off-site backups of validator state, alert routing
- **$500** — Public bug bounty pool, paid on verified findings in the PQ boundary
- **$500** — Ops and upgrade budget: extra agent runs, emergency hosts, counsel hours when needed

## What the lean path requires from the operator

1. **Re-scope the audit gate.** The spec currently says no mainnet without a clean external audit. The lean path launches mainnet permissioned, with all five validators Cymetica-run and only platform-custodied value on the chain, and moves the external audit to the gate that admits outside validators or third-party builders at scale. One line in the spec's gate table.
2. **Defer counsel to the economics decision.** ET10 as the native gas asset changes the token's treasury and regulatory profile. The F0 gate's public-communication rules stay in force meanwhile: no presale, no "mainnet" dates in public copy before F3, testnet tokens described as valueless.
3. **Accept a longer road to decentralisation.** Outside validators come after the audit, 90 days of measured liveness, and a reproducible node a third party can run unaided. On the lean path that is a revenue-funded milestone, not a launch-day feature.

## What does not change

Engineering gates stay as written:

- NIST ACVP known-answer tests green on every merge for both precompiles
- Differential tests against an independent implementation on every PQ change
- Two independent builders produce byte-identical release binaries
- Signing-latency and gas-pricing benchmarks pass before testnet
- Chaos gate on testnet: deliberate double-sign, liveness fault, key rotation, emergency upgrade

The chain itself:

- CometBFT + Cosmos SDK v0.50 with the EVM module, so Solidity builders arrive with their existing tooling
- SLH-DSA-128s validator keys; ML-DSA-44 accounts with hybrid and PQ-native transaction envelopes
- 2-second blocks, five equal-power genesis validators, one may be down without loss of liveness
- Validators never expose RPC; public endpoints are separate sentry-fronted full nodes
- Custody rules unchanged: user funds stay segregated, nothing moves from a deposit wallet

## Timeline on the lean path

- **Weeks 1–2 — Stage 1 live.** Provenance rails on Ethereum, first revenue line for the ET10 base.
- **Weeks 2–10 — F1 devnet.** Chain builds, PQ gates green, five validators on private hosts.
- **Weeks 10–16 — F2 public testnet.** Explorer, faucet, SDK; chaos gate; legal gate re-run on the actual copy.
- **Weeks 16–22 — F3 + F4.** Permissioned mainnet, ET10 bridge, EventTrader rail live. Bounty opens.

About five months to a running mainnet. The full-price path is six to twelve months, because the audit engagement and counsel package sit on the critical path.

## Read it plainly

The lean path is cheap because the expensive items are assurance, not construction, and assurance can be staged when the only value on the chain is ours. It is not free of risk: until the external audit is bought, the post-quantum boundary is verified by our own gates and a public bounty, not by an outside firm. That is a fair trade for a permissioned chain carrying platform-custodied value, and the wrong trade for one carrying strangers' money. The plan keeps those two states separate.