Financial projections

The plan — not bookings

Every revenue cell is a plan. There are no named customers and no cash already raised.

What this is. The planning model for the $20 million seed. Every revenue cell is a plan. There are no named customers, no awarded government contracts, and no cash already raised.

What this is not. A CRM export. A share of the $58.5 billion defense AI request. A developer self-serve funnel.

How to read it. Enterprise clients back the launch by sitting a proof, signing, and expanding. Traction is logos and booked annual contract value — not waitlists, press, or inbound curiosity.

Source of the arithmetic: institutional enterprise-contract model (S2). Same numbers as the investor business plan, spelled out for diligence.


The traction path

The product launch is not a website. It is the first CIO who lets us install the archive on their hardware.

  1. Seed builds the factory. $20 million funds a repeatable 6–16 week proof, Fortune 100 hygiene, an air-gap path, and $2 million in customer-side credits.
  2. A named committee sits the proof. NDA, then CORE / DIVISION / ENTERPRISE on their floor. The credit means they are not funding a science experiment.
  3. They sign. They own the memory. That logo is traction. Three logos or $15 million booked is the gate that says the market will pay.
  4. They expand. More divisions, more seats, more of the archive. Planning net revenue retention is 140 percent — Palantir printed 157 percent; we have not run this cohort yet.
  5. Those logos back Series A. The trigger is $80 million recognized run-rate, or fifteen logos plus one government task order of $15 million or more. Public brand spend waits until then.

If step 3 misses, the company bridges. It does not pivot into awareness.


What we sell

Proposed price book. Not live.

SKUAnnual contractProofWho it is for
SSI-CORE$2 million6 weeks; $150k creditableFirst land, one division
SSI-DIVISION$8 million10 weeks; $400k creditableA line of business
SSI-ENTERPRISE$25 million16 weeks; $1.0M creditableFlagship or sovereign install

Planning bands as logos accumulate: commercial $4–12 million, sovereign $15–40 million. Those bands are Year-2 and Year-3 expanded value at the comparables. We are not underwriting two $40 million flags in Year 1.


Year-1 mix — how eight customers become $48 million

Assumptions. Funding closes. Eight-week hire and proof sprint. First signatures in month 5. Year 1 is the twelve months after close. Deals signed after month 6 recognize 50 percent of booked annual value; deals signed by month 6 recognize 75 percent. Base mix is six commercial and two government. Blended recognition in the base is 50 percent.

Base — $96 million booked / $48 million recognized

Logo typeCountAnnual valueBookedRecognized in Year 1
Commercial DIVISION-class4$8M$32M$16M
Commercial upper band2$12M$24M$12M
Government vehicle (OTA / task order)1$15M$15M$8M
Sovereign / flagship1$25M$25M$12M
Total8$96M$48M

The $15 million and $25 million cells are a stretch inside base. They assume a government vehicle and a flagship land in the same year as the first commercial installs.

Downside — $20 million booked / $10 million recognized

Three $4 million commercial lands plus one $8 million government vehicle. Four logos. Phase 0 barely clears the gate — or misses it.

Upside — $176 million booked / $80 million recognized

Eight $12 million commercial tickets plus two $40 million sovereign tickets. Ten logos. Sensitivity only. Not the underwriting case.

Year-1 recognized headline: $10 million / $48 million / $80 million (downside / base / upside).

Conversion analog used to get into the room: 40 percent on CORE, 40 percent on DIVISION, 35 percent on ENTERPRISE. That band is from early Palantir-style pilots. Trinity has not run these proofs as a measured cohort.


Five-year plan

Growth after Year 1 assumes 140 percent net revenue retention, eight / twelve / sixteen / twelve net-new logos in Years 2–5, and a mix shift toward $12 million and $25 million tickets as references exist. Years 4–5 are base only.

Dollars in millions. A positive cash number is cash consumed. A number in parentheses is cash generated.

MetricY1 downY1 baseY1 upY2 baseY3 baseY4 baseY5 base
Logos (year end)481018324862
Booked ACV2096176180380620980
Recognized revenue1048801403205801,000
Gross margin58%62%65%68%72%74%76%
Gross profit5.829.852.095.2230429760
R&D8121622405570
Sales and marketing6101428456075
G&A35710162024
EBITDA(11.2)2.815.035.2129294591
EBITDA margin6%19%25%40%51%59%
Cash consume / (generate)148(5)(20)(100)(250)(520)
Heads28425595160210260

How to read Year 1 base. $48 million recognized × 62 percent gross margin = $29.8 million gross profit. Subtract R&D $12 million, sales $10 million, G&A $5 million = EBITDA $2.8 million (6 percent). Sales expense is named-account work and proofs — not a brand campaign.

How to read Years 2–5. Booked annual value leads recognized revenue because multi-year enterprise contracts recognize over the service period. Year 5 booked $980 million versus recognized $1.0 billion as prior-year backlog converts.

Base Year 3 ($320 million, 32 logos) sits inside a $250–400 million planning band. Base Year 5 ($1.0 billion, 62 logos) sits inside $800 million–$1.2 billion.

Year-1 base consumes $8 million of the $20 million ask. Downside consumes $14 million. The ask is sized so a miss is a bridge, not a media pivot.


What must be true at the same time

ScenarioWhat has to be true together
Downside4 logos. CORE-class $4 million tickets. One $8 million government vehicle. Expansion muted. Federal path may stall at SBIR Phase I ($323,090). Public brand stays off. Series A does not trip.
BaseFunding closes. 8-week sprint. 8 logos (6 commercial + 2 government). $96 million booked / $48 million recognized. 140% NRR. Phase 0 gate clears in months 0–9.
Upside10 Year-1 logos. Eight $12 million commercial tickets. Two $40 million sovereign tickets. Cash generative in Year 1.

Unit economics (base plan)

Commercial first customerSovereign / flagship
Annual value (blended)$8 million$20 million
Cost to win$0.9 million (two AEs + proof + legal)$2.4 million (sales + facility + counsel)
Five-year LTV (140% NRR, 8% churn)$28 million$72 million
LTV : CAC31×30×
Payback5–8 months8–12 months

Those ratios collapse if $10 million-class conversations do not convert. C3.ai FY2026 revenue was down 36 percent versus FY2025 — that is the failure mode: many large conversations, zero signatures.


The invoice we take, not the market we invent

Trinity takes budget the CIO already defends.

ViewFigureWhat it is
Typical 10,000-employee AI stack$4–8 million / yearSeats, tokens, integrator retainers
Typical 100,000-employee estate$30–55 million / yearMulti-division production
Fortune 1–50 displaceable slice$4.5 billion (range $2.0–8.0B)A pool. Not our revenue.
Modeled Fortune 50 three-year takeaway$90.9 millionWhat they pay the rented stack minus what they would pay to own the memory

Palantir closed 73 deals of $10 million or more in Q2 2026. That print is existence of a buyer who writes this check. It is not our pipeline.

The Department of War’s $58.5 billion AI request is a policy line. It is not our serviceable market. First federal cash, if any, is SBIR Phase I at a $323,090 cap, or a live prototype other-transaction. Allied 2026–28 SSI opportunity ($90–140 million) is not a fourth revenue column.


What would break the base case

  1. No logo in nine months. Public brand stays off. The raise becomes a bridge.
  2. Federal path stuck at SBIR Phase I. Government Year-1 recognized cash collapses below $1 million.
  3. An incumbent ships owned persistent memory into the air-gap. The takeaway story weakens.
  4. C3-style non-conversion. Many $10 million conversations, zero signatures.

Series A — what the first customers unlock

RoundAmountTrigger
Seed$20 million askThis document
Series A$250 million ($200–300 million band)$80 million recognized run-rate or 15 logos plus one government task order of $15 million or more

Use of funds for this seed: ../02-capital/COMPANY-UOF-20M.md. Contract gates stay in the private data room.

The first enterprise clients do not just buy a product. They install the archive, expand it, and become the proof that the next raise is a scale raise — not another seed.

$20 million is an ask. No customer logos yet. Forecasts are a plan.