Part IX. Company and Investing · Chapter 82
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.
In this chapter 9 sections
A defensible memo ties a dated decision to a buyer problem, stack position, primary evidence, technical dependencies, baseline, economics, scenarios, kill criteria, and the next observable; its conclusion must recompute when those inputs change.
A memo is valuable when a skeptical colleague can change an input and reproduce the new decision. Narrative polish is secondary. Facts, company claims, analyst inferences, and scenarios occupy different fields, and every current assumption has an owner and expiry. An aggregate score would hide which one failed.
The memo is an executable decision record
The two-page architecture contains decision and scope; buyer and repeated problem; controlled stack layer; primary evidence; technical gates and baseline; customer and economic evidence; dependencies and financing; downside, base and upside scenarios without invented probabilities; kill criteria; conflicts; and next evidence event.
Two-page evidence architecture
GAO and the National Academies provide technology context. Randomized benchmarking and the benchmarking review delimit performance evidence. ACM and official interface documents support rerunnable technical checks. None provides company-specific buyer, revenue, cost, financing, or rights evidence; those must come from primary company, contract, filing, and customer records in a real memo.
| Claim under review | Source chain |
|---|---|
| Technology assessments provide dated state-of-technology evidence but do not substitute for company primary materials or customer verification. | |
| Benchmark results must preserve comparator and system boundary before entering a commercial thesis. | |
| Reproducible artifacts and documented interfaces make technical diligence rerunnable. | |
| A technical memo should separate established algorithmic results from company execution and market inferences. |
Scenario table without a synthetic valuation
| Decision input | Base record | Kill condition | Owner |
|---|---|---|---|
| Backend portability | 2 adapters pass task | fewer than 2 | technical lead |
| Buyer renewal | 3 of 4 | below 3 | commercial lead |
| Review age | 0 days | over 90 days | memo owner |
| Decision | conditional proceed | any kill true | committee |
The sample developer-platform memo uses two decision gates: cross-backend portability and a buyer renewal threshold. Both are observable. The base record passes and returns conditional proceed. Removing portability or falling below the renewal threshold changes the label to reject without editing the prose. Downside and upside remain sensitivity cases, not forecasts with decorative probabilities.
METRICS = {"portable_backends": "backends", "renewals": "buyers",
"eligible_buyers": "buyers", "age": "days"}
def valid_iso(value):
parts = value.split("-") if type(value) is str else []
return len(parts) == 3 and [len(part) for part in parts] == [4, 2, 2] and all(part.isdigit() for part in parts)
def decide(memo, maximum_age_days=90):
errors = []
if memo.get("scenario") not in {"base", "downside", "upside"} or memo.get("claim_kind") not in {"fact", "inference"}:
errors.append("scenario_or_claim_kind")
sources = memo.get("sources", {})
if type(sources) is not dict or not sources:
errors.append("sources")
else:
for source_id, source_date in sources.items():
if not valid_iso(source_date):
errors.append(source_id + "_date")
for name, unit in METRICS.items():
item = memo.get(name, {})
if type(item.get("value")) is not int or item.get("unit") != unit:
errors.append(name + "_schema")
failures = []
if not errors:
if memo["portable_backends"]["value"] < 2:
failures.append("portability")
if memo["renewals"]["value"] < 3:
failures.append("renewals")
if memo["age"]["value"] > maximum_age_days:
failures.append("freshness")
return {"decision": "invalid" if errors else ("reject" if failures else "conditional-proceed"),
"errors": sorted(set(errors)), "failed_gates": failures}
memo = {"scenario": "base", "claim_kind": "inference",
"sources": {"scenario:renewal-ledger": "2026-08-01", "scenario:portability-run": "2026-08-10"},
"portable_backends": {"value": 2, "unit": "backends"},
"renewals": {"value": 3, "unit": "buyers"},
"eligible_buyers": {"value": 4, "unit": "buyers"}, "age": {"value": 4, "unit": "days"}}
base_result = decide(memo)
bad_result = decide({**memo, "sources": {"scenario:renewal-ledger": ""}})
renewal_result = decide({**memo, "scenario": "downside", "renewals": {"value": 2, "unit": "buyers"}})
portability_result = decide({**memo, "scenario": "downside", "portable_backends": {"value": 1, "unit": "backends"}})
assert base_result["decision"] == "conditional-proceed"
assert bad_result["decision"] == "invalid" and "scenario:renewal-ledger_date" in bad_result["errors"]
assert renewal_result["failed_gates"] == ["renewals"] and portability_result["failed_gates"] == ["portability"]
print(f"PASS: 82 memo workbook base={base_result['decision']} invalid={bad_result['errors']} renewal={renewal_result['decision']} portability={portability_result['decision']}")Artifact contract. A dated compact memo dataset and executable scenario/expiry evaluator. Every conclusion traces to an assumption or source; stale assumptions fail; base, downside, and upside remain scenarios rather than invented probabilities; kill criteria can reverse the label.
Exact validation command: python3 tools/validate_briefs.py --briefs data/editorial_briefs_64_87.json --from 64 --through 87 --check-rewritten-sources --execute-artifacts
Assumptions get owners and expiry dates
Assign every assumption a reviewer and date. Put source links at the claim, record contrary evidence, and show the effect of each sensitivity. The conclusion then names maximum commitment, the artifact to obtain, and the two conditions that end the thesis. This makes a short memo denser and more useful than a long market narrative.
Write a memo another reviewer can rerun
Prompt. Write a two-page quantum-company decision memo using only dated, attributable evidence.
Deliverable. Decision, buyer, stack, evidence ledger, baseline, dependencies, three unweighted scenarios, two kill criteria, owner, and ninety-day next review.
Pass condition. Every material fact has a source and date, inference is labeled, no scenario is presented as a forecast probability, and either kill criterion changes the recommendation when triggered.
Model answer: conditional proceed with two kill criteria
Format. Conditional-proceed memo for a developer platform with portability and buyer-renewal gates.
The model answer conditionally proceeds with a bounded diligence partnership. Portability passes two adapters and buyer renewal meets the declared threshold. Either failure flips the evaluator to reject. The memo labels technical evidence separately from commercial inference, lists conflicts, and expires in ninety days. It contains no valuation because the available record does not justify one.
Verification. The evaluator returns proceed under the base inputs and reject when either the portability test or renewal threshold fails.
Companion work
Artifacts for this chapter
These entries resolve to checked-in local source. Commands are reproduced exactly from the chapter manifest, and source-embedded fixtures are exported as direct downloads.
diligence dossier
Evidence-linked memo decision record
Reproduce or test
python3 tools/validate_briefs.py --briefs data/editorial_briefs_64_87.json --from 64 --through 87 --check-rewritten-sources --execute-artifacts
Provenance
Sources and review
- U.S. Government Accountability Office. Quantum Computing and Communications: Status and Prospects. GAO. 2021government technology assessment
- Association for Computing Machinery. Artifact Review and Badging. ACM Publications. 2026official reproducibility policy
- OpenQASM Technical Steering Committee. OpenQASM 3 specification. Linux Foundation Joint Development Foundation. 2026official technical specification
- IBM Quantum. Qiskit documentation. IBM. 2026official documentation
- Google Quantum AI. Cirq documentation. Google. 2026official documentation
- Easwar Magesan, J. M. Gambetta, and Joseph Emerson. Scalable and robust randomized benchmarking of quantum processes. Physical Review Letters. 2011primary paper
- Timothy Proctor et al.. Benchmarking quantum computers. Nature Reviews Physics. 2025peer-reviewed perspective
- National Academies of Sciences, Engineering, and Medicine. Quantum Computing: Progress and Prospects. National Academies Press. 2019consensus study report
The load-bearing claims in the chapter are mapped inline to this registered source set. A citation supports only the bounded claim beside it.