← Blog Raise catalog Team Capital / $20M Ask

Use of funds

Eighty-five percent builds the factory that puts memory on their floor.

Eighty-five percent builds the factory that puts memory on their floor.

Where the check goes

Trinity Sky is asking for a $20 million seed. The figure is an ask — not cash in hand, not a closed round. The product is a runtime the buyer installs on-prem. The raise stands the people and the path that can put that runtime in a named committee’s building in six, ten, or sixteen weeks, keep the memory stack maintainable, and convert the first proofs into owned installs. Those installs become the book that unlocks Series A.

Eighty-five percent of the check is four jobs a later associate can name without a glossary: product, deploy, evidence, and customer-side proof credits. Five percent is the named-account desk. Two and a half percent is reserve. This raise does not buy a public campaign.

Use of funds is a stock. The Year-1 operating plan is a period. On the base path the company consumes $8 million of cash at 42 heads. On the downside path it consumes $14 million. The $48 million recognized cell is the plan if the first logos exist. Spending the ask does not print that cell.

Eighty-five percent of the raise is product, deploy, evidence, and customer-side proof credits.

The eight-line envelope

The book and the capital memo print the same eight lines. They are the envelope. They are exact. Each amount appears once.

  • $7.0M factory. Engineers who can run a six-to-sixteen-week proof and keep the memory stack maintainable. The thing being sold is a runtime the buyer puts on-prem.
  • $3.0M air-gap. Trinity’s own disconnected install path, so a proof never sends the buyer’s data through a public cloud. Customer GPU rooms and SCIFs stay the customer’s capital.
  • $3.0M enterprise. Single sign-on, SIEM, access control, restore — Fortune 100 hygiene. The difference between a demo and a renewable contract.
  • $2.0M validation. An independent party re-runs the recall clocks and writes the threat model. Isolated recall on a quiet machine is 13.834 µs. Under load the same stack has returned 5,127 µs. Those are different experiments. We show both.
  • $2.0M credits. Customer-side proof credits so a CIO opens the proof on an operating line. First line we stop if the gate misses.
  • $1.5M IP. Export, assignment, and custody hygiene — required before a government task order is discussable.
  • $1.0M sales. Two account executives and legal. Fully loaded commercial cost to open a named account is about $0.9 million; this line covers that desk. The rest of Year-1 selling sits in operating expense once the company is converting.
  • $0.5M reserve. Thin operating reserve. Entity, books, insurance. Intentionally thin.

A partner meeting can follow the money in one pass. We are not bidding on the $58.5 billion defense AI request. Air-gap is Trinity’s own disconnected path, not a claim on a policy line.

Product, deploy, evidence, credits

Product is the bind-and-unbind work that keeps a memory stack shippable. A six-to-sixteen-week proof is a motion only if the same engineers can run it twice. Deploy is two paths that share one job: the archive never leaves the buyer’s perimeter. One path is a disconnected install, so a proof does not route the buyer’s data through a public cloud. The other is Fortune 100 hygiene — single sign-on, SIEM, access control, restore. Without those, a proof is a demo. With them, it can become an annual contract.

Evidence is not an internal slide. It is an independent party re-running the recall clocks and writing a threat model a Series A associate can hold. Isolated recall and loaded recall are different experiments; both belong in the folder. Credits sit on the customer side of the table so a CIO opens the proof on an operating line. That is how a takeaway sale starts.

Those four jobs are why eighty-five percent of the check looks the way it does. Five percent is two account executives plus legal — enough to open a named account, not enough to pretend the raise is a national campaign.

A proof they can own

The buyer is a Fortune 100 or government procurement committee. They sit a proof of value with their data, on hardware they already control. A committee signs an NDA. Then a six-, ten-, or sixteen-week proof installs the archive, runs the recall protocol, and shows fail-closed behavior when the system is not sure. The proof fee is credited against the land. They sign. They own it.

They write a check for an archive they own. Their install is the product launch.

CORE, DIVISION, and ENTERPRISE are proposed SKUs in the book — $2 million, $8 million, and $25 million annual — and the install stays in their building. Proof length tracks the land: about six weeks for a first division, ten for a line of business, sixteen when air-gap is the default path. Proposed proof credits against those lands are $150,000, $400,000, and one million dollars. Commercial first customers sit in a $4–12 million band. Year 1 is the first lands, not two $40 million flags.

Their expansion is the traction. Those logos, when they exist, are the Series A story. Seed exists so the first proofs can earn them.

What seed must produce

The gate that says the raise worked is three signed logos or $15 million booked. Either prong is enough. Waitlists, press, and inbound curiosity stay off the scoreboard. Hit the gate, and the company is a first-customer business with installs on customer hardware. Miss it, and the company revises the sales plan. If Phase 0 fails, the credits line is the first line we stop.

Series A is a $250 million target in a $200–300 million size band after either $80 million recognized run-rate, or fifteen logos plus one government task order of $15 million or more. That band is raise size, not seed pre-money. Seed enables the trigger. The envelope stays on product, deploy, evidence, and credits until the trigger exists. Year-1 base booked of $96 million is the operating plan if those outputs exist. Downside recognized is $10 million. None of those cells is a guaranteed result of spending the raise.

$20 million is an ask. There are no customer logos yet. Forecasts are a plan. Unused seed does not become a street campaign; it stays as runway.

Sources: docs/gtms/09-raise/BUSINESS-PLAN-INVESTOR.md §13 — The $20 million. Line-by-line rationale: docs/gtms/09-raise/02-capital/COMPANY-UOF-20M.md. Isolated 13.834 µs and loaded 5,127 µs clocks restated from the same book §2 when the validation line is named. $20 million is an ask. No named logos.

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