Investment Memo and Startup Thesis Templates
Most quantum investment write-ups are pitches wearing a memo's clothes. This chapter gives you a template that forces the opposite shape: decision first, uncertainty near the top, and proof gates you can put on a calendar.
The quality of an investment memo is measured by how cheaply it lets you change your mind. A memo that cannot name the evidence that would reverse its conclusion is marketing with footnotes.
This chapter gives you the working template: a two-page structure that ties build, partner, invest, monitor, wait, and avoid labels to source-labeled evidence, proof gates, and kill criteria — useful to founders writing a thesis and to investors testing one. Everything here is educational framing, not financial advice.
Core concepts: company diligence, startup wedge selection, investment proof gates.
A memo is a decision, written down
An investment memo or startup thesis exists to support one of six labels: build, partner, invest, monitor, wait, or avoid. Everything in the document either moves that label or is decoration. The memo states the decision up front, then the company or wedge, the customer and their problem, the technical claim, the evidence for it with source quality labeled, the market timing, the risks, the proof gates ahead, the kill criteria, and — the part most drafts skip — what new evidence would change the conclusion.
Its job in this book is to make your reasoning inspectable. If another engineer reads it, they should see exactly which claim carries the decision and which evidence carries the claim. A memo that survives that reading is a decision document; one that does not is a pitch with better typography.
Two formulas that keep the memo honest
The first scores the opportunity: . It forces probability, value, and timing onto one line, and it punishes the common habit of treating waiting as free.
The second scores the decision itself: . A company can hold a beautiful claim and still score poorly, because nothing has been proven yet and the kill criteria are close.
Neither number is precise, and precision is not the point. The point is that changing any input visibly moves the output, so the memo's logic can be argued with instead of admired.
Anatomy of a two-page memo
Two pages, ten sections, in this order:
- Decision label
- Company or wedge
- Customer and problem
- Technical claim
- Source-labeled evidence
- Baseline or alternative
- Moat and timing
- Risks and kill criteria
- Proof gates
- Next evidence to gather
A memo built this way can be wrong later and still be good work, because it recorded what would change the answer. The proof gate at the end is a measurable event — a reproduced benchmark, a signed pilot, an error rate published with its method — never a mood.
How memos go wrong
The most common failure is the pitch in memo clothing: certainty up front, caveats in the appendix. A diligence memo runs the other way — uncertainty near the top, where the reader can price it.
Two more failures show up constantly. One is mixing technical and market claims without showing the dependency: "the market is huge" and "the device works" are separate claims that fail independently. The other is score theater — a weighted rubric with no source labels, which lends subjective enthusiasm the visual grammar of rigor.
Finally, many memos hide the alternative. Every decision has a baseline: use the classical tool, wait for hardware, partner with a lab, buy a component, do nothing. A memo that never names what happens instead has not made a decision; it has expressed a preference.
The engineer's version
If you write software, you already know this document. It is an architecture decision record: context, decision, alternatives considered, consequences, assumptions, and the triggers that reopen the question. Decision records are built to be revisited, and so is this memo.
Quantum makes the revisit loop non-optional. Roadmaps slip, error rates improve, and last year's kill criterion becomes this year's marketing slide. A source-refresh loop — dated claims, scheduled re-checks — keeps the memo alive without making it twitch at every headline.
What this buys a builder or an investor
The same template serves three readers. A founder gets a startup thesis and an MVP test plan. An investor or partner gets an evidence review they can audit in ten minutes. A learner gets an artifact that proves they can connect physics, software, applications, and timing — worth more than another certificate.
Keep the template short. Length hides weak assumptions, and the two-page limit is a feature. The best memo makes the decision, the risks, and the next evidence easy to find and easy to attack.
Exercise
Write the memo you would want to receive. Choose one quantum company or one startup wedge and write a two-page memo with an explicit decision label, risks, proof gates, and kill criteria.
- Submit: the memo, naming the three proof gates that would raise your confidence and the evidence that would reverse the decision.
- Check: compute both scores — opportunity EV and the decision score — and write one sentence on which input you trust least.
- Repair: if the draft reads like a pitch or cites evidence without source labels, rework it after Chapter 81, Reading Roadmaps Like an Operator, which shows how to turn roadmap language into testable claims.
Check your understanding
Without looking at the template: what are the ten sections, and why does the decision label come first?
A passing answer explains that the label disciplines everything below it — the evidence, gates, and kill criteria exist to move or defend that label. Oral defense: take a real company roadmap and argue both the invest and the avoid memo from the same facts.
If you get stuck
If your memo hides its uncertainty, confuses a funding announcement with a technical result, or cannot state what would change its mind, go back to Chapter 81 (Reading Roadmaps Like an Operator) and Chapter 72 (When to Build, Partner, Wait, or Avoid). The first fixes evidence translation; the second fixes the decision labels themselves.