CSAI Business Plan

The fork

Closed USD, closed DFY, or a listed print

Three architectures. All keep the print settled by identified nodes. They disagree about the currency of the bond and the fee, about whether the network is a customer list, and about what a Q4 2026 tokenization harbor is allowed to change. Present values use the v0.18 method on the current gate and multiple. Bonded-DFY is scored with the harbor on — the most generous reading of that path. Listing run-rates are working figures, labelled as such.

Present value on these levers

USD thousands, discounted three years. DFY includes the 35% exclusion haircut. V19 is look-through (equity + NetCo + W at fundamentals).

Bonded · USD

$10.4M

SKU mix of v0.18

Bonded · DFY

$5.3M

Haircut 35% · harbor on

Listed · open

$11.6M

Methodology licences

V19 look-through

$12.9M

Includes 15% W at fundamental

	Bonded · USD	Bonded · DFY	Listed · open
Print settlement	Identified 4-of-5, USD bond floor (I1, I2)	Identified 4-of-5; DFY bond breaks I2 unless a residual USD floor is kept	Same 4-of-5 as integrity, not as a SKU. USD floor kept.
Primary revenue	Freeze / desk / Read induced by admission	Same SKU mix, paid in DFY	Listing / methodology licence. Freeze residual.
Token	None (W, if any, earned and untouched)	DFY as rail. 86.8% insider. Not W.	Token only supports listings: DFY as wrapper unit, already live
Q4 2026 harbor	Not needed	Assumed. Covers wrappers, not the rail, not the insider test, not the USD floor	Helps wrappers list. Does not require CSAI to touch DFY.
Fourteen lines	14 hold	6 break (1, 2, 3, 9, 11, 14)	14 hold
Publisher exclusion	Clean, with a franchise-shaped Freeze licence to explain	Lost. CSAI’s efforts drive DFY by construction.	Cleanest. Impersonal listed print at a USD list.
VASP / MSB	None (vasp_v0.17)	High. Accepting DFY for fees; DFY bond custody.	None. Venue’s listing, not CSAI’s rail.
Investable object	Publisher. Network is demand.	Publisher + DFY mark. Two theses, one cap table.	Publisher only. Network non-investable.
Live fit, 11 Sep 2026	Bond pool is USD. Zero named firms. Flywheel has not started.	Wrapper /DFY is live. Bond and fees are not. This column is a conversion, not a description.	Matches /invest and /network. F2 listed now. Readers free.
What the method will not count	A Freeze flywheel with no named firms	A DFY premium, a harbor that has not been enacted, an insider holding marked to market	MutantDeFi / FloatNet current feed revenue (still unquantified)

Harbor column is an assumption from 1 October 2026, not enacted text. Bonded-DFY is scored with the harbor on, which is the most generous reading of that path.

Admission is the product

Closed bonded · USD

Sponsored products need five identified, bonded signers. Each product induces Freeze licences, desk licences and Reads, all in USD. The publisher is paid by the population the network admits. This is v0.17/v0.18 as written.

Customer: Sponsors (DCMs, prediction books, perp venues) and the nodes they require.
SKU: Freeze $24k, desk $24k, Read $12k, settlement royalty max($25k, 0.5 bp). Year-3 one product ≈ $219k to CSAI.
Network: Closed. Identity + USD bond + Freeze licence. One node, one vote. Open readers unpaid.
Token: None. W, if it exists, is earned, never sold, 0% insider, never held by CSAI.
Harbor: Not required. USD licences and a USD bond do not wait on a tokenization harbor.
Investable: The publisher, on MSCI / CF Benchmarks comparables. Network is not a security.
First money: Week 23 settlement licence (plan). Requires five named firms at the week-12 gate.
Live: Lab seats, not named firms. Bond pool is already USD. No Freeze licence sold. The flywheel has not started.
Refuses: Token in the quorum; bond in DFY or W; CSAI as operator or fee collector.

The rail is the token that already exists

Closed bonded · DFY

Same closed admission network, with DFY as the bond and fee currency, under an assumed Q4 2026 tokenization safe harbor. The harbor is taken as given for this column. It is not in the v0.17–v0.19 record, and it is not law.

Customer: Same sponsors and nodes, now paying and posting in DFY.
SKU: The v0.18 SKU mix, denominated in DFY at a published DFY list. CSAI then prices, accepts and references DFY.
Network: Closed. Identity + DFY bond (USD floor dropped, or residual USD with DFY above it). Same 4-of-5.
Token: DFY. Issuer-connected holding 86.8%. Not W. Cannot pass the 20% insider-supply test. Clean-history W was designed specifically so this token would not have to.
Harbor: Assumed available 1 October 2026. Working reading: it covers listed tokenized wrappers (Genus I/II, the /DFY units already on the tape). It does not convert a DFY bond into a USD floor, does not cure 86.8% insider, and does not let CSAI accept DFY for the Read without ending the publisher exclusion.
Investable: Publisher plus a DFY mark. Two theses, one cap table — the mess entity_structure_v0.19 removed. Comparables no longer apply without a conglomerate discount.
First money: Same gate, plus a DFY market that must exist for bonds to have a mark on draw day.
Live: Instruments are already quoted /DFY. The bond, the clerk ticket and NAV are USD. Live DFY is a wrapper numeraire, not this rail.
Refuses: Nothing the fourteen lines were written to refuse — which is why this column scores like P5.

Plan of record

Listed · open

A non-investable distributed network. Signers are integrity a regulator can ask for, not a customer cohort. Readers are free. Any token exists only to support listings — DFY as the wrapper numeraire of GM! Core, or W as a work record the Read never prices. Primary revenue is the listing / methodology licence.

Customer: Venues and protocols that list or fund against F2 / P3 / P4 — not the nodes.
SKU: Listing / methodology licence at a USD list; Read as research data; royalty on listed notional. Freeze collapses to a handful of named signers.
Network: Non-investable. Open readers, no name, no bond. Bonded 4-of-5 remains as integrity, not as SKU. USD bond floor kept.
Token: Only as listing support. DFY = unit of the listed wrapper (already live). W = optional unpaid work record. Neither is sold by CSAI. Neither is a fee rail.
Harbor: Helps, if it lands: tokenized wrappers (notes, funds, perps, prediction contracts) list against a signed print without each venue writing a formula. Matches the /invest sentence exactly. Does not require CSAI to touch DFY.
Investable: The publisher only. No NetCo, no DFY round, no network round. Cleanest publisher exclusion. MSCI / CF Benchmarks apply without a sentence of explanation.
First money: F2 as perp funding mark — already live on ETH + Canton. P3 dated wing 8 February 2027.
Live: This is what /network and /invest currently are: “listed now”, “perp funding mark”, “wrappers can list and settle on any venue”.
Refuses: Selling admission. Selling DFY or W from CSAI. Making the network the investable object.

Two instruments, cover as a product

V19 NetCo (parked)

A network company sells W once as a security to fund the open layer, holds a royalty share, and later operates a backstop vault. CSAI holds 15% of W and 35% of NetCo. Five lines amended. Not a DFY programme.

Customer: Accredited buyers of W; sponsors who want cover; open nodes paid in W.
SKU: NetCo equity + W. Publisher lines unchanged except D63’s 20% royalty share.
Network: Still I1–I7. Token never in the quorum. Vault pays after the bond.
Token: W, sold once as a security, no profit right, so it can migrate if CLARITY passes. Not DFY.
Harbor: Irrelevant to formation. W is sold as a security before enactment; the harbor does not change that.
Investable: Two offerings, two disclosures. Adds ~$3M of CSAI look-through at fundamentals, 8% cover effect.
First money: Week-12 NetCo formation; week-16 sale; week-23 vault. Nothing the live site is doing.
Live: Absent. Taking it now would introduce a token into a product that has none.
Refuses: P5. Selling W from CSAI. Putting economics in W instead of in NetCo equity. Substituting DFY for W.

The harbor, stated as an assumption

Assumed available 1 October 2026. As-of 11 September 2026, that is a scenario switch, not a statute. CLARITY cloture is 15 September. Nothing in the calendar waits on either.

What it would cover

Listed tokenized wrappers against a signed print (Genus I dated strip; Genus II perpetual freeze).
The /invest sentence: notes, funds, tokens, perps, prediction contracts listing without each venue writing a formula.
Possibly a limited offering or distribution of a tokenized unit, if the enacted text says so — counsel, not this pack.

What it would not

Convert a DFY bond into a USD floor (I2).
Cure an issuer-connected 86.8% holding against a 20% insider-supply test.
Let CSAI price, accept or reference DFY for the Read, Freeze or royalty without ending the publisher exclusion.
Make DFY into W. W was designed so the token that already exists would not have to fit.
Move the 23 November gate or the 8 February money line. Nothing in the calendar waits on it.

Seven invariants — not on the table

fluxnet_v0.19 §1. Bonded-DFY is the path that asks to move I2. The answer in this pack is no: other collateral sits above the USD floor, never in place of it. A harbor does not rewrite that sentence.

I1
Print is the quorum value of identified registered nodes under the frozen rule.
I2
Bond has a USD floor. Other collateral sits above it, never in place of it.
I3
No token is a quorum input, a settlement value or a fallback.
I4
Governance is one registered node, one vote, on identity. No governance through any token.
I5
Registry is immutable, no admin key. CSAI holds no key and does not deploy.
I6
Fallback order never polls a dealer. IOSCO statement attaches to every settlement / listing licence.
I7
Sponsors and venues pay counterparties directly. No entity collects a fee to pay a node.

Comparables — they are listing businesses

MSCI

EV/EBITDA 22.7× (3 Sep 2026); Index segment margin 75.6%; Index run-rate $1.9bn.

Mature index businesses are licensed underliers, not bonded networks. Run-rate is the metric.

A 6,000-person public company. Not a private-company multiple.

Cboe Global Indices / VIX

The incumbent second-moment primitive. Listed futures (VX), securities options, notes, a trademark.

The book’s replacement thesis is a listing thesis. Revenue is methodology + listed complex, not node admission.

Incumbent owns the rulebook, the exchange, and the trademark. A successor either migrates inside or fights.

CF Benchmarks

FCA-authorised crypto benchmark administrator; acquired by Kraken 2019, nine-figure press report; 2022 rev. ~£6m.

Closest like-for-like. A regulated administrator, not a network token. Sits inside an exchange group.

Strategic buyer who needed it. Directional, not a filing.

The v0.18 comparable set was already an index-administrator set. Using it to price a bonded node-admission business is the conglomerate discount entity_structure_v0.19 wrote down. Using it to price a DFY rail is worse: the comparables do not hold tokens. Using it to price a listed underlier is what the method was built for. Live F2 as a funding mark is the first observation that the comparable set and the product finally match.

The desk is not the gate

Lona (long-bias) and Shae (short-bias) already pay the same F2 on ETH Sepolia and Canton, cycle 79. Instruments are /DFY. Tickets are USD. That is a counterparty cycle on a listed wrapper. It is not the 4-of-5 signer gate, it is not a desk licence, and it is not evidence that a DFY bond rail has started. Treating the lab books as either would be a category error the raise cannot survive.