BillionToOne, Inc. [BLLN] — Company Research (Task 1)

Resolved identity (verified before any analysis was performed):

This is a molecular diagnostics company, not a fuel-cell or onsite-power company. It was resolved strictly on evidence. The near-matches flagged in the brief were checked and rejected: Ballard Power Systems is BLDP (CIK 1453015) and Bloom Energy is BE (CIK 1664703). No other ticker in the SEC file begins with BLL. BLLN sits in the med-tech cohort, alongside this book's NTRA / GH / TXG / TWST coverage — not the Bloom Energy cohort.

1. Mention-frequency FIRST — run generatively, before any view was formed

Per references/mention-frequency.md, this was executed before the model, the valuation, or any directional view.

Corpus (ONE source, named and pinned): EDGAR 8-K Exhibit 99.1 earnings releases. Three quarters — 2025Q3 (filed 2025-12-09), 2025Q4 (2026-03-04), 2026Q1 (2026-05-06). Word counts 1,908 / 2,210 / 2,630. All counts are normalised per 10,000 words.

Window statement — stated plainly rather than papered over. Three quarters is the entire post-IPO history of this company; it is not a 12–16 quarter series and no claim here should be read as one. Alpha Vantage (the skill's primary transcript source) is exhausted for the day. I attempted to build a longer, richer series from Motley Fool call transcripts and only Q1 2026 exists there (Q3-25 and Q4-25 return HTTP 404), so a three-point transcript series from a single source could not be constructed. Rather than mix sources within a series — the exact error references/mention-frequency.md records from the SMR re-run — the series stays on the EX-99.1 corpus, and the Q1 2026 transcript is used separately and labelled as such for qualitative colour only.

Honest verdict on the metric for this name: it is not yet informative. Earnings press releases are formulaic and short; across three points the apparent movements are dominated by the product-launch news cycle rather than by management attention. Specifically, Northstar reads 0 → 27.1 → 0 per 10k words, which is not a decay signal — it reflects Q4-25 being the release that announced Northstar PGx and Northstar Select CH. operating income reads 21.0 → 22.6 → 0 purely because the Q1-26 release used the phrase "Income from operations". Both are artifacts. Neither is reported as a finding.

The one durable observation across all three releases is that guidance rises (5.2 → 18.1 → 15.2) alongside two successive guidance raises — consistent with, but not independent of, the raises themselves.

Open questions generated by this pass (no interpretation attached at the time they were listed):

1. Why does fetal cell appear for the first time in 2026Q1 (11.4/10k)? → Investigated: Unity Confirm, launched 1 May 2026, a circulating-fetal-cell confirmation assay. Corroborated in the 8-K and the 10-Q.
2. Why does contracted lives appear for the first time in 2026Q1? → Investigated: the Anthem contract took contracted lives to ~300m (>90% of the US population). This turned out to be the single most economically important item in the quarter, and the mention-frequency pass is what surfaced it.
3. Why is ASP mentioned in every release but never with a stated ceiling? → Investigated: this became the central analytical question of the memo (Section 4).

Item 2 is the one genuine generative hit: the payor-contracting vector was found by asking why a phrase appeared, not by looking for evidence of a prior view.

2. The business

BillionToOne runs a single proprietary platform — Quantitative Counting Templates (QCT), a single-molecule next-generation-sequencing (smNGS) method that counts individual cell-free DNA molecules — and sells two product families off it.

Prenatal (UNITY) — 89.1% of Q1-2026 revenue ($96.5m). UNITY was the first single-gene non-invasive prenatal test (sgNIPT), assessing fetal risk for recessive conditions (sickle cell, cystic fibrosis) from a maternal blood draw without a paternal sample or an invasive procedure. It has since broadened to aneuploidy, fetal antigen and carrier screening from one draw. In 2024 the fetal-antigen tests drove national medical-guideline changes. The company states it tests approximately 1 in 11 US pregnancies. Recent additions: expanded Red Blood Cell and first-and-only Platelet Fetal Antigen NIPTs (Feb 2026), and Unity Confirm (May 2026), a fetal-cell-based confirmation assay for high-risk screens.

Oncology (Northstar) — 9.9% of Q1-2026 revenue ($10.7m), growing 392% YoY. Northstar Select (therapy selection; claims 2–5x lower limit of detection and >50% more actionable mutations than comparators) and Northstar Response (methylation-based tumour-burden monitoring). Northstar Select obtained MolDX/Medicare coverage under LCD L38043 in 2025. Northstar PGx and Northstar Select CH launched Q1 2026. A tumour-naive MRD test is guided to launch by end-2026.

Clinical trial support — 1.1%. Immaterial.

Geography: substantially all revenue is United States (10-K, verbatim). There is no international business to model. Charts 04 and 05 are the mandatory product and geography exhibits.

3. Financial trajectory — tied to filed statements

Source: FY2025 10-K "Results of operations" table and Q1-2026 10-Q Statements of Operations. Every line above was read off the filed statement and is reproduced exactly by the Excel model — see the tie-out block on the Income Statement tab (all six differences compute to 0.00).

FY2025 was the first profitable year. Volume: 616,000 tests accessioned in FY2025 (410,000 FY2024; 272,000 FY2023); 610,000 delivered and billable. Overall ASP rose 35% in FY2025 and 28% YoY in Q1-2026 to $571. COGS per test was $153 in Q1-2026, down 5% sequentially.

Balance sheet: $537.5m cash at 3/31/26 against a $90.0m term loan (carried at fair value, 8% coupon) — net cash ~$447m. Operating lease liability $50.8m. There is no financing risk and no refinancing wall.

4. The analytical crux: how much of the growth is the ASP engine, and is it finite?

Growth decomposes as volume × price. In Q1-2026: volume +44%, ASP +28%. The ASP component is driven by payor contracting — moving from out-of-network to in-network status, which the 10-K notes both raises realised price and makes the sales motion easier. Contracted lives went from >250m (FY2025 10-K) to ~300m after the Anthem contract, which management describes as >90% of the US population.

That engine is, by arithmetic, finite. Once ~90%+ of covered lives are contracted, further ASP growth must come from guideline-driven mix, new higher-priced products, or oncology reimbursement — not from re-contracting the same population. This is the single most important modelling question on the name, and it is why the memo does not simply extrapolate 84% growth.

4a. The prior-period revenue true-up — a real earnings-quality finding

The 10-Q revenue note discloses revenue recognised for performance obligations satisfied in prior periods (retroactive re-pricing of previously-billed claims when new payor agreements are signed):

Q1-2026's single-quarter true-up exceeded the whole of FY2025, and equals 8.5% of the quarter's revenue. Stripping it (and allocating FY2025's $8.7m as $2.9m in Q1-25 per the filing plus an assumed even $1.93m/quarter across Q2–Q4-25 — this quarterly allocation is an assumption, not disclosed):

Sensitivity on the assumption: if Q4-25's true-up were $0, organic sequential is +3.3%; if Q4-25 absorbed all $5.8m, it is +9.9%. Under every allocation, organic sequential growth decelerated sharply from the +15% to +26% range that prevailed through 2025. The conclusion is robust to the assumption; the precise number is not. See chart 03 and chart 23.

This is not an accounting abuse — it is cash the company is genuinely owed and the disclosure is clear and voluntary at note level. But it flatters the headline, and management confirmed on the Q1 call that "guidance for true-up revenue includes only historical Q1 levels; no additional true-up revenue is embedded in projections for subsequent quarters." So the guide is not being propped up by this line going forward — but Q1's optics were.

5. Ownership, insiders and control

Dual class. As of 4 May 2026: 41,442,834 Class A (1 vote) and 4,552,650 Class B (15 votes) = 45,995,484 shares outstanding. Class B holds ~62% of voting power on ~9.9% of the economics. Public holders cannot effect change of control.

Insider activity — 39 Form 4s parsed directly from EDGAR XML (2025-11-10 to 2026-07-20):

• Open-market purchases (code P): $3.18m — every one at exactly $60.00 on 10–12 November 2025. These are IPO-price participations (director Thomas Bremner $2.50m; Nancy Johnson $0.41m; others smaller), not opportunistic buying. There has been no open-market insider purchase at any post-IPO market price.
• Open-market sales (code S): $21.62m, all between 2026-06-10 and 2026-07-20 — i.e. entirely after the 180-day lock-up expired on 2026-05-05. CEO Oguzhan Atay $10.10m; David Tsao $4.54m; Nancy Johnson $3.32m; Shan Sakakibara $2.67m; John ten Bosch $0.99m. Twenty-one Form 144s were filed over the same window.

Applying the empirical asymmetry (Cohen, Malloy & Pomorski 2012) honestly: insider sales are mostly noise, and post-lock-up diversification by founders and early employees of a nine-month-old IPO is the most routine category of sale there is. I do not treat this selling as bearish evidence. The more informative observation is the absence of the signal that does carry weight: no clustered, non-routine, multi-insider buying at market prices. That is a neutral-to-mildly-negative read, not a bearish one. See chart 21.

Institutional ownership: three Schedule 13Gs and one 13D are on file. EDGAR provides no single endpoint returning all holders of a ticker, so this is explicitly a partial pull rather than a complete ownership picture.

6. Internal control — a real, unremediated weakness

The FY2025 10-K discloses material weaknesses in internal control over financial reporting that remain unremediated:

1. An insufficient complement of resources with appropriate technical accounting knowledge, and an inability to consistently establish segregation of duties in finance and accounting;
2. Controls over the analysis, accounting and presentation of certain financial instruments (common stock warrants; term-loan presentation);
3. IT general controls — segregation of duties and privileged access over journal-entry creation and posting.

Remediation is "ongoing in 2026" (third-party technical accounting support, additional hires).

Scoping this correctly matters and I want to be precise: the identified weaknesses concern financial instruments and ITGC — they do not name revenue recognition, which is the load-bearing estimate on this company. That materially limits how much weight this can carry. But revenue here is heavily estimate-dependent (variable consideration, expected-value method, hindsight adjustments), the true-up line is accelerating, and the control environment is by the company's own admission under-resourced. The honest characterisation is: a real yellow flag that does not currently reach the revenue line, on a company where the revenue line is unusually estimate-dependent.

7. Competitive position and the evidence base

The oncology evidence gap is the most important competitive fact on this name. A PubMed query on 2026-07-28 returns 151 publications for Signatera and 1 for Northstar Select liquid biopsy. A ClinicalTrials.gov sponsor query returns only three BillionToOne-sponsored studies: the ADVANCE prenatal study (NCT07643896, N=1,000, recruiting, started 2026-01-10, primary completion 2027-12) and two small investigator-scale ctDNA oncology trials (NCT07096362, N=50, pancreatic; NCT05733689, N=20, gastro-oesophageal). There is no registered MRD trial, and management stated on the Q1 call that MRD supporting data will be "published concurrently at launch rather than beforehand."

I opened these records and read the fields rather than citing them second-hand — the failure mode references/mention-frequency.md records from the ISRG pass. See chart 18.

Read: the prenatal franchise is real, differentiated, guideline-endorsed and defensible. The oncology franchise is early, thinly evidenced, and about to enter the single most evidence-intensive competitive arena in diagnostics against an incumbent with a 151-to-1 publication advantage. That asymmetry is the crux of Section 4 of the valuation work.

8. TAM — bottom-up, per references/tam-sizing.md

Built from units. No industry-report headline is used.

Prenatal (US only — the company has no international business):

• TAM (100% of births screened, BLLN captures all): 3.6m × 1.7 × $600 ≈ $3.7bn. Not a real target.
• SAM (80% penetration, a generous 25% BLLN share): 3.6m × 0.80 × 0.25 × 1.7 × $600 ≈ $735m.
• BLLN FY2025 prenatal revenue $277.1m ⇒ ~38% of that SAM already captured.

Oncology: ~1.9m new US cancer cases/year; the advanced/metastatic solid-tumour pool addressable by therapy-selection liquid biopsy is roughly 600–700k patients (estimate). At a Medicare-anchored ~$3,000 therapy-selection ASP and 20% penetration, that is ~$400m; an MRD/monitoring franchise is a multiple of that on repeat testing, which is precisely why Signatera is Natera's largest business. BLLN's oncology revenue is $10.7m in Q1-2026 ($42.9m annualised).

Time to revenue: MRD launches end-2026 at the earliest; Northstar Response Medicare coverage is targeted for year-end 2026; the ADVANCE study primary completion is 2027-12. Nothing here contributes materially to FY2026, and only modestly to FY2027.

Penetration-path sensitivity is carried in the Bear/Base/Bull revenue paths (chart 25) rather than as a separate table, since the paths are the penetration scenarios.

Required implied-penetration statement (see valuation §5): at $133.64, a reverse-DCF requires FY2035 revenue of ~$3.5bn — 11.5x FY2025. Prenatal at the generous SAM above tops out near $735–850m, so the current price requires oncology to become roughly three-quarters of the company — a business today carrying $42.9m of annualised revenue, one peer-reviewed publication, no registered MRD trial, and no Medicare coverage for its monitoring assay.

9. Risks

1. ASP ceiling. ~300m contracted lives is >90% of the US population; the re-contracting engine is close to spent.
2. Oncology execution against Signatera with a 151-to-1 publication deficit and no registered MRD trial.
3. Reimbursement/regulatory. MolDX coverage decisions are discretionary; FDA LDT oversight is unsettled.
4. Revenue estimation risk. Variable-consideration accounting with hindsight adjustment, an accelerating true-up line, and unremediated ICFR material weaknesses in adjacent areas.
5. Concentration. ~89% prenatal, ~100% US — no diversification whatsoever.
6. Post-lock-up supply. Lock-up expired 2026-05-05; $21.6m sold in six weeks and 21 Form 144s filed.
7. Dual-class control. Class B carries 15 votes/share; no governance recourse.
8. Competitive response. Natera and Guardant are both larger and better capitalised in oncology.