Phase A — Understand the business
Lens 1 · Company Overview
Tempus AI is a precision-medicine company that runs on a two-sided flywheel: it performs molecular diagnostic testing (mostly oncology — sequencing tumors and now, post-Ambry, inherited-cancer/hereditary genetics), and in doing so accumulates one of the largest multimodal clinical-genomic data libraries in the world (8M+ de-identified patient records per ), which it then licenses to pharma and biotech for drug discovery and trial design. Tests feed the data library; the data library is what makes the tests differentiated and what pharma pays for. Management frames the whole thing as "AI applied to precision medicine," but — see Lens 13 — the genuinely AI-native revenue line is still tiny.
Two reported product lines:
- Diagnostics — clinical oncology NGS panels (xT, xF liquid biopsy, xM MRD/monitoring), plus hereditary testing (Ambry, acquired Feb 2025). FY2025 revenue $955.4M, +111% YoY.
- Data and applications — de-identified data licensing to pharma (the AstraZeneca/Pathos/SoftBank deals live here) plus software (Lens, Notebooks, the "Insights" line). FY2025 revenue $316.4M, +31% YoY.
Contract structure / payment terms. Diagnostics is a reimbursement business — revenue depends on Medicare/Medicaid and commercial-payor coverage and rates per test, not contracted recurring fees; volume × reimbursement is the engine. Data is lumpy, multi-year licensing — e.g. the AstraZeneca/Pathos deal is a $200M data-licensing + model-development arrangement, recognized over time, which makes the data line spiky quarter to quarter. Product concentration is real even if customer concentration isn't: clinical-oncology + hereditary tests together were 74% of total revenue in FY2025 (up from 63% in FY2024 and FY2023), and "No single customer represents a material portion of the Company's accounts receivable". So the risk isn't one whale customer churning — it's reimbursement policy and oncology test volume.
Listed on Nasdaq as TEM (Class A, $0.0001 par); IPO'd June 2024. Headquartered in Chicago; founder-CEO Eric Lefkofsky.
Lens 2 · Supply Chain
Tempus sits in the middle of a lab-services + data value chain. Named stakeholders along the chain:
Upstream (inputs into Tempus):
- Sequencing instruments & consumables → Illumina (the dominant NGS platform; Tempus is an Illumina customer for sequencers/reagents — a single-source-ish dependency shared by the entire clinical-genomics industry). Oxford Nanopore / Pacific Biosciences are secondary long-read options.
- Lab reagents, oligos, flow cells — Thermo Fisher, Agilent, and similar life-science suppliers.
- Tissue & sample inflow → hospitals, academic medical centers, community oncology practices — these are simultaneously suppliers of samples and customers of results. ~50%+ of US academic cancer centers are connected per company materials.
- Cloud / compute → hyperscaler(s) for the "cloud-and-compute" environment behind Lens/Notebooks (AWS/GCP-class infrastructure; the AI-model work is compute-hungry).
- Reference-lab capability → now internalized via Ambry (Aliso Viejo, CA), which was previously Tempus's external hereditary reference lab and is now owned.
Downstream (who pays Tempus):
- Payors — CMS/Medicare + Medicaid (the marginal price-setter for diagnostics), commercial insurers. This is the true revenue chokepoint for the Diagnostics line.
- Ordering clinicians / oncologists — the demand generators (they order the test).
- Pharma & biotech data buyers — AstraZeneca, Pathos AI, GSK, Merck (Merck collaboration cited as backing 2026 guidance ), plus the SB TEMPUS JV with SoftBank for the Japan market.
Chokepoints / single-source dependencies: (1) Illumina — instrument/consumable dependency for the core sequencing workflow; (2) CMS reimbursement — a policy change (e.g. enforcement of the Medicare 14-Day Rule, see Lens 10) directly hits Diagnostics revenue; (3) de-identified data supply is self-generated — the moat input (patient records) comes from Tempus's own test volume, so the supply chain is partly internal and self-reinforcing, which is the bull case for the data flywheel.
Lens 3 · Competitive Advantages (moats)
The durable moat, if there is one, is the multimodal data library + network effects, not the assays:
- Data scale / proprietary corpus (real moat). 8M+ de-identified multimodal records (genomics + clinical notes + imaging + transcriptomics) linked longitudinally. Competitors have genomic data; few have it linked to clinical outcomes at this scale. Ambry was bought partly for its hereditary-genetics database, explicitly to train models — Tempus treats acquisitions as data acquisitions. This is the asset pharma pays for and the hardest thing for a new entrant to replicate (you'd need years of test volume across connected health systems).
- Network effects (genuine but slow). More connected oncologists → more tests → richer data → more attractive to pharma → funds more tests/products. Tempus describes its two product lines as designed to "enable and enhance the other". This is the strongest part of the thesis.
- Switching costs (moderate). Once a hospital's oncology workflow, EMR integration, and ordering habits route through Tempus, displacement is sticky — but not locked; Guardant, Natera, Foundation Medicine all compete test-by-test.
- Brand / "AI" positioning (contested). The "AI" brand is a marketing moat with the equity market more than a technical one — see Lens 13. Perceived value is high among growth investors; the demonstrated AI-revenue is low.
Bargaining power: Weak vs. payors (CMS sets the price; Tempus is a price-taker on reimbursement). Weak vs. Illumina (instrument dependency). Improving vs. pharma data buyers (the bigger and more unique the corpus, the more leverage — but pharma can also walk, and the AstraZeneca/Pathos deals were flagged by shorts as "weakening," see Lens 13). Net: the moat is the data corpus, and it's real; the pricing power around it is not yet proven outside a handful of marquee deals.
Lens 4 · Segments
Two product segments; no separate geographic segment disclosure of materiality (US-dominant; Japan via the SB TEMPUS JV is nascent). All /:
| Segment | FY2024 rev | FY2025 rev | YoY | Q1-2026 rev | Q1 YoY |
|---|
| Diagnostics | $451.7M | $955.4M | +111% | $261.1M | +35% |
| Data and applications | $241.6M | $316.4M | +31% | $87.0M | +41% |
| Total net revenue | $693.4M | $1,271.8M | +83% | $348.1M | +36% |
Trend & cause:
- Diagnostics +111% in FY2025 is mostly inorganic — the Ambry acquisition added ~$362.7M of Diagnostics revenue and pushed hereditary tests to ~460,500 in the year. Strip Ambry and underlying oncology volume grew ~26-29%. So the headline 83% total growth flatters the organic rate; the clean organic growth is closer to the mid-30s% — which is exactly what Q1-2026 (+36% total) shows now that Ambry is in the base.
- Data +31% FY2025 → +41% in Q1-2026 is the more interesting signal: the data line is re-accelerating, driven by the AstraZeneca/Pathos/Merck licensing deals and "Insights" (+44% in Q1). The data line is the higher-margin, more-moat-defining business; its reacceleration is the part of the story bulls should care about most.
- MRD (minimal residual disease) — small base, explosive growth: ~6,500 tests in Q1-2026, +~500% YoY. This is the future oncology-monitoring TAM and the locus of the Guardant patent fight (Lens 10).
Gross-margin by line (FY2025):
- Diagnostics: (955.4 − 386.1)/955.4 = ~59.6%
- Data & applications: (316.4 − 87.8)/316.4 = ~72.3%
- Blended: (1,271.8 − 473.9)/1,271.8 = ~62.7% (up from ~55% in FY2024 ). Mix-shift + scale is lifting gross margin ~8 pts YoY — a real positive.
Phase B — Measure performance
Lens 5 · Earnings Result (latest print: Q1 2026, reported 2026-05-05)
All figures unless:
- Revenue $348.1M, +36.1% YoY (vs $255.7M Q1-2025). Beat on the top line. Diagnostics $261.1M (+35%), Data $87.0M (+41%).
- Loss from operations $(84.7)M, worse than Q1-2025's $(68.7)M — opex (incl. a step-up in SBC) outran the gross-profit gain.
- Net loss $(125.9)M vs $(68.0)M — the much larger net loss is below the operating line: interest expense on the post-Ambry debt + other items. The market called this an "EPS miss" despite the revenue beat.
- Adjusted EBITDA $(2.8)M vs $(16.2)M — marching toward breakeven on the adjusted metric. (Note the shorts' objection to what's adjusted out — Lens 10/13.)
- Gross margin improving: diagnostics COGS $101.0M on $261.1M rev (~61.3% GM); data COGS $25.1M on $87.0M (~71.1% GM).
- Balance-sheet flags: Cash $521.2M (down from $604.8M at YE2025 — ~$83M total-cash draw in the quarter); company cites $643.8M cash + marketable securities at 3/31/26. Operating cash used $(73.3)M (improved from $(105.6)M Q1-2025). Long-term debt ~$204.6M. Accumulated deficit deepened to $(2,521.8)M.
- SBC re-accelerated to $52.7M (vs $23.0M Q1-2025) — a ~2.3× jump that is doing real work in the "adjusted" numbers (see Lens 10).
- Guidance RAISED: FY2026 revenue $1.59–1.60B (~+25–26% over FY2025's $1.27B) and FY2026 Adjusted EBITDA ~+$65M — the first positive full-year adjusted-EBITDA guide, framed as backed by the Merck collaboration.
- Market reaction: stock has been weak into mid-2026 (~$50–52 in June vs a $63.54 June-2024 reference and higher 2025 peaks ) — the tape is rewarding neither the revenue beat nor the EBITDA guide, consistent with a market that is pricing credibility/forensic risk, not growth risk.
Unusual vs. own history: operating loss widened YoY while adjusted EBITDA improved — the gap is entirely SBC + below-the-line interest. That divergence (GAAP deteriorating, adjusted improving) is the single most important thing to watch.
Lens 6 · Earnings Calls (sentiment trend)
No transcripts on the research shelf (transcripts=0), so this is ``-grounded across the last several quarters:
- Management's consistent focus: "AI-enabled precision medicine," the data-library scale, the data-licensing pipeline (AstraZeneca → Pathos → Merck), and a path to positive adjusted EBITDA. The 2026 guide makes "EBITDA breakeven/positive" the headline KPI — a deliberate pivot from a pure growth story to a "growth and discipline" story, which is what loss-making 2024-IPO names have to do once the market stops paying for growth alone.
- Tone shift (3-4 call arc): from land-grab / TAM-expansion (2024 post-IPO) → narrative defense (mid-2025, after the Spruce Point report, see Lens 13) → prove-the-margin (2026, leaning on the raised EBITDA guide and the Merck deal as validation). The recurring phrases are "multimodal," "foundation model," "operating leverage," and increasingly "adjusted EBITDA." The thing they now have to keep saying — because the market is skeptical — is that the data deals are real, recurring, and arms-length.
- What to listen for next call: organic (ex-Ambry) Diagnostics volume, Data-line renewal (not just new signings), MRD ramp, and any change in how aggressively adjusted EBITDA is defined.
Lens 7 · Comps
Peer set = US precision-oncology diagnostics + clinical-genomics data.
| Company | Ticker | Mkt cap | EV/Sales | P/E | Notes |
|---|
| Tempus AI | TEM | ~$9–9.5B | ~5.6× FY25 rev, ~5.8× FY26E | n/m (loss-making) | 36% organic growth, adj-EBITDA ~breakeven |
| Guardant Health | GH | ~$4.5B | P/S ~12–14× TTM | n/m | liquid-biopsy leader; suing TEM |
| Natera | NTRA | n/a this run | n/a | n/m → approaching profitability | MRD (Signatera) leader |
| Exact Sciences | EXAS | acquired by Abbott $21B, closed 2026-03-23 | ~3× TTM sales pre-deal; deal ~$21B on ~$3.25B FY25 rev ≈ ~6.5× sales | n/m | screening (Cologuard); now Abbott |
| Foundation Medicine | — | private (Roche-owned) | n/a — not public | n/a | tissue CGP; Roche subsidiary |
| Caris Life Sciences | CAI | n/a this run | n/a | n/m | IPO'd 2025; closest data-play comp |
| Illumina | ILMN | n/a this run | n/a | positive | supplier, not pure comp |
Read: TEM at ~5.6× EV/forward-sales is cheaper than Guardant (~12–14× P/S) on a multiple basis, but Guardant carries a near-pure liquid-biopsy growth story while TEM's print is half lower-margin diagnostics. The Exact Sciences/Abbott $21B take-out at ~6.5× sales is the most useful anchor: it says a strategic will pay mid-single-digit sales multiples for a profitable-ish US oncology-diagnostics franchise — which both supports TEM's floor and feeds the recurring "is TEM itself a take-out target?" chatter. Do not over-anchor: Natera/Caris/Illumina multiples were not sourced this run — flagged n/a rather than guessed.
Lens 8 · Stock-Price Catalysts (moves >5%, ~last 2 years since IPO)
TEM has only traded since June 2024, so the "5-year" window is really "since IPO." Pattern is catalyst-driven and violent, ``:
- Jun 2024 — IPO at $37, popped ~+9% day one.
- Apr 2025 — AstraZeneca/Pathos $200M deal announced → sharp rally ("Tempus AI Surges on Landmark Partnerships"). The market reacts hard to data-licensing deal headlines — they validate the moat thesis.
- May 28 2025 — Spruce Point "Strong Sell" report → stock −16–19% to ~$54.80 intraday. The single biggest down-catalyst: a forensic/governance shock, not an operational one.
- Quarterly earnings — repeatedly whippy on the adjusted-EBITDA trajectory and guidance revisions more than on the revenue line (the Q1-2026 "beat revenue, miss EPS" is the template).
- 2026 — Exact Sciences/Abbott $21B deal reframed sector M&A and put TEM on take-out-speculation watch.
What the market actually reacts to: (1) data-deal headlines (up), (2) forensic/governance credibility (down — Spruce Point), (3) the path-to-profit / EBITDA guide (more than revenue), (4) sector M&A. It does not reward raw revenue growth at this point — growth is assumed; the debate is quality and trust.
Phase C — Judge people & books
Lens 9 · Management
- Eric Lefkofsky — Founder & CEO. Serial entrepreneur: co-founder of Groupon, Echo Global Logistics, InnerWorkings, Mediaocean; founded Tempus in 2015. Builder archetype, unambiguously — he starts and scales companies.
- Track record — mixed-to-poor for public shareholders. This is the crux. Groupon IPO'd at ~$13B (2011) and was a disaster post-IPO; Lefkofsky was CEO Aug 2013–Nov 2015 through its struggles. InnerWorkings sold to PE for a fraction of its IPO market cap (2021). Echo Global and InnerWorkings "did well for Lefkofsky" per but the pattern shorts cite is: promote a disruptive-tech narrative, monetize early, public holders left with lackluster-to-negative returns. He is, in Spruce Point's framing, a "cash-out king." Fair or not, that reputation is in the price and is a structural overhang on the multiple.
- Skin in the game & control. Large insider — and crucially, super-voting Class B at 30 votes per share. That's far above the 10–20 votes/share that's already considered aggressive. Lefkofsky effectively controls Tempus regardless of economic ownership. Pro: founder conviction, long-horizon bets (Ambry). Con: minority shareholders have no governance recourse — related-party deals (Pathos), comp, and strategy are his call. For a company whose chief risk is related-party / accounting trust, a 30-vote founder is the worst possible governance backdrop.
- Capital allocation. Aggressive M&A (Ambry $600M, debt-funded), large data-deal structuring, heavy SBC. ROE/ROIC are negative (loss-making, accumulated deficit $(2.5)B) so there's no value-creation track record at Tempus yet to judge — the bet is that the data flywheel eventually earns a return on all this reinvestment.
- Red flags: (1) the 30-vote structure; (2) related-party adjacency — Pathos AI is a Lefkofsky-linked entity and a counterparty in the flagship AstraZeneca data deal; (3) the SoftBank JV optics (Lens 10); (4) a CEO whose prior public companies underperformed. Founder archetype + this stage usually argues for patience — but here the governance + related-party combination argues for a trust discount until the cash flow is undeniable.
Lens 10 · Forensic Red Flags
Acting as a forensic analyst. Several genuine yellow-to-orange flags; none yet a smoking gun, but the cluster is why the multiple is capped. All numeric items unless.
- GAAP-vs-adjusted divergence (the central flag). FY2025 operating loss narrowed to $(252.9)M from $(691.1)M, and Adjusted EBITDA improved to $(7.4)M from $(104.7)M — but operating cash flow got worse: $(218.1)M used vs $(189.0)M in FY2024. Earnings quality is moving the opposite way to the adjusted metric the company is selling. Watch this above all else.
- Receivables outrunning revenue. AR grew to $311.2M from $154.8M (+101%) while revenue grew +83%. Some is Ambry consolidation, but AR-growth > revenue-growth is the classic place where aggressive revenue recognition or collection risk hides. Allowance only $2.8M.
- Customers paying in equity. Tempus received $32M of marketable equity securities on accounts receivable in FY2025 (and $22M in each of 2024/2023). Booking revenue collected in private/illiquid equity of customers is a real quality-of-revenue question and a valuation-mark risk.
- SBC as the adjustment lever. SBC was $534.1M in FY2024 (IPO catch-up), $124.7M FY2025, then re-accelerated to $52.7M in Q1-2026 alone (vs $23.0M). The entire "approaching breakeven" story rests on adding SBC back; the Q1 re-acceleration means dilution is a live, growing cost the adjusted number hides.
- "Novel debt-forgiveness structure" in adjusted EBITDA — Spruce Point alleges the adjusted-EBITDA calc benefits from a debt-forgiveness mechanic. Corroborating signal on the shelf: FY2025 shows a $12.0M loss on debt extinguishment and a refinancing (proceeds $196M, repaid $276.9M, $7.8M prepay premium) — i.e. active, complex debt maneuvering around the Ambry financing.
- SoftBank JV round-trip optics (the most-cited item). A SoftBank affiliate put $200M into Tempus's pre-IPO Series G-5 preferred on Apr 30 2024; weeks later (Jun 27 2024) the SB TEMPUS JV — each side contributing
¥15.0B ($95M) — let Tempus recognize $95M of TCV, exactly equal to its own contribution. The appearance of capital being routed in a circle to manufacture revenue/income is the single hardest allegation to dismiss; Tempus calls the report "riddled with hypotheticals and inaccuracies".
- Goodwill/intangibles ballooned to $470.2M / $355.3M (from $73.3M / $11.7M) post-Ambry — future impairment risk if Ambry underperforms. Plus a $51.7M tax benefit in FY2025 (vs a small provision prior) that flattered the net-loss line — Ambry deferred-tax-driven, non-operational.
Regulatory findings (required sub-section):
- SEC: No Litigation Releases and no AAERs name Tempus AI in 2021-06-18 → 2026-06-18.
- DOJ / U.S. Attorney (EDNY): Civil Investigative Demand received Mar 4 2024 re: False Claims Act / Anti-Kickback statute / the Medicare 14-Day (Date-of-Service) Rule — Tempus is producing documents on a rolling basis; the 10-K explicitly says "no assurance … that it will not result in a material adverse effect". This is the most material live regulatory exposure — it goes to the legitimacy of Diagnostics reimbursement. Also an Ohio AG subpoena (2022, resolved).
- Litigation: Guardant Health patent suit (filed Jun 11 2024, D. Del.) — alleges Tempus xF, xF+, xM Monitor, xM MRD infringe five Guardant liquid-biopsy patents; seeks injunction + enhanced damages + royalty. Plus a Guardant declaratory-judgment action (Jan 17 2025) on advertising claims. Genetic-data class actions over the Ambry data-disclosure / collection of genetic information.
- Short-seller forensic opinion: Spruce Point "Strong Sell," est. −50% to −60% long-term downside.
- Verdict: clean at the SEC-enforcement level; not clean on the operating/forensic and litigation level — the DOJ CID (reimbursement integrity), the Guardant patent fight (touches the fastest-growing MRD line), and the cluster of accounting-quality flags are each individually survivable but collectively material.
Phase D — Project & stress-test
Lens 11 · Forward Projection (EPS, next three fiscal years — FY2026/27/28)
Bottom-up from FY2025 actuals + the FY2026 guide. Output ``, inputs labeled. Tempus is loss-making, so the scoreable line is path to positive EPS, not a near-term EPS level.
Revenue:
- FY2026: $1.595B (midpoint of the $1.59–1.60B guide).
- FY2027: ~$1.99B.
- FY2028: ~$2.43B.
Margin / profit path:
- Blended gross margin drifts from ~63% (FY25) toward ~66–68% as the higher-margin Data line mixes up.
- Adjusted EBITDA: FY2026 ~+$65M (guide) → FY2027 ~+$180M → FY2028 ~+$320M.
- GAAP EPS: FY2026 still a net loss ~$(1.20)–(1.50)/sh; GAAP breakeven not until ~FY2028–2029 once SBC normalizes and interest is covered. The gap between "adjusted-EBITDA positive in 2026" and "GAAP-EPS positive ~2028+" is the investment debate.
Base / Bull / Bear (FY2027 adj. EBITDA frame):
- Base: revenue ~$1.99B, adj-EBITDA ~$180M, GAAP still slightly negative. Data line renews and grows; no reimbursement shock.
- Bull: Data line compounds 40%+ on more AZ/Merck-style deals, MRD scales, adj-EBITDA ~$250M+, market re-rates to a "profitable data company" multiple.
- Bear: DOJ reimbursement action and/or Guardant injunction clip Diagnostics; data deals prove lumpy/non-recurring; SBC keeps GAAP losses fat; multiple compresses to the Spruce Point −50% zone.
(Per --watchlist rules: no forecast.ts create logged in the breadth loop.)
Lens 12 · Bull vs Bear
Bull case. Tempus owns a structurally scarce asset — the largest linked multimodal oncology data corpus in the US — and is monetizing it two ways (diagnostics reimbursement + pharma data licensing) with the higher-margin data line re-accelerating to +41%. Organic growth is a real mid-30s%, gross margin is climbing ~8 pts/yr, adjusted EBITDA turns positive in 2026 (guided), and the Ambry deal both bought a profitable hereditary-testing franchise and a training-data trove. Marquee counterparties (AstraZeneca, Merck, GSK, SoftBank) validate the data thesis. The Exact/Abbott $21B take-out proves strategics pay up for US oncology-diagnostics scale — TEM at ~5.6× forward sales is cheaper than Guardant and is itself plausible M&A. If the flywheel works, this is a compounding healthcare-data platform, not a lab.
Bear case (2–3 permanent-impairment risks). (1) Reimbursement is the whole Diagnostics business and it's under a DOJ/Anti-Kickback/14-Day-Rule investigation — an adverse outcome doesn't dent a quarter, it questions the legitimacy of how the largest revenue line gets paid. (2) The "AI" is barely monetized (~$12.4M/2% of 2024 revenue per Spruce Point) — if pharma data deals prove lumpy and non-recurring rather than a durable subscription stream, the moat doesn't translate to pricing power and the multiple de-rates to a lab-services multiple. (3) Trust/governance — round-trip-flavored related-party structures (SoftBank, Pathos) under a 30-vote founder with a cash-out reputation means the market will never pay full multiple until cash flow is undeniable; the discount is structural, not sentiment.
Pre-mortem (18 months out, thesis broke): The DOJ matter escalated or a Guardant injunction hit the MRD/liquid-biopsy line; a couple of pharma data deals weren't renewed, exposing the data revenue as project-based not recurring; SBC kept GAAP losses wide while cash burned; a short follow-up re-litigated the round-trip allegations; the stock sat at the Spruce Point downside.
Are multiples too high? On sales, no — ~5.6× forward is reasonable-to-cheap for 25–36% growth. On trust-adjusted earnings, the market is right to discount: there are no GAAP earnings, cash flow is negative, and the quality flags are real. The multiple isn't expensive; the certainty is.
Contrarian view (what the market refuses to see): The Data-and-applications line, not Diagnostics, is the company — and it's the part the market under-models because it's lumpy. If even two or three more AZ/Merck-scale deals convert into renewing multi-year licenses, TEM re-rates as a software/data company and today's diagnostics-multiple framing looks far too low. The bull case is less "AI" and more "the only company that can sell linked oncology outcomes data at scale."
Lens 13 · Devil's Advocate (short-seller)
Dismantling the bull case.
- The brand is a tell. They renamed from Tempus Labs to Tempus AI right into the AI mania, yet AI-native applications were ~2% of 2024 revenue. This is a diagnostics lab with a data side-business wearing an AI costume. Re-rate it as the lab it mostly is and the stock is worth a lot less.
- Revenue concentration is in the least-defensible place: 74% of revenue is clinical-oncology + hereditary testing — reimbursement-dependent, and there's a DOJ CID on exactly that (Anti-Kickback, 14-Day Rule). If CMS/DOJ tightens, the core revenue mechanism is impaired, not just slowed.
- The data deals may be circular. The SoftBank JV recognized $95M TCV equal to Tempus's own $95M contribution, weeks after a SoftBank affiliate funded $200M of pre-IPO preferred. Pathos AI — counterparty in the flagship AstraZeneca deal — is Lefkofsky-linked. If the marquee "validation" deals are partly related-party capital dressed as third-party revenue, the entire moat narrative is suspect.
- Most dangerous competitor bulls underestimate: Guardant (suing over the MRD/liquid-biopsy IP — and MRD is Tempus's headline growth vector at +500%) and Natera (Signatera is the MRD category leader and is actually approaching profitability). Tempus is behind in the one area it's touting as its future, and is being sued there.
- Worst capital-allocation / incentive flags: 30-vote super-voting control; related-party Pathos; SBC that re-accelerated 2.3× in Q1-2026 even as management sells a "disciplined / breakeven" story; debt-funded M&A creating $825M of goodwill+intangibles to impair.
- Assumptions that must hold for today's price: data licensing is recurring not project-lumpy; reimbursement survives DOJ scrutiny intact; Guardant loses or settles cheaply; SBC normalizes; adjusted EBITDA converts to real cash flow (it hasn't — OCF got worse).
- If growth disappoints 20–30%: with no GAAP earnings, the stock has only a sales multiple to stand on; a decel + a credibility event = the −50% Spruce Point scenario is the base, not the tail.
- Single scenario that permanently impairs: an adverse DOJ/False-Claims resolution that forces a reimbursement-practice change across the Diagnostics book — that's existential for the majority of revenue, not cyclical.
Lens 14 · Management Questions (ordered by information value)
- The DOJ EDNY Civil Investigative Demand on the Anti-Kickback statute and the Medicare 14-Day Rule — what specific billing practices are under review, and what % of Diagnostics revenue depends on the practice in question?
- Of FY2025 Data-and-applications revenue, how much is recurring/renewing multi-year licensing vs. one-time upfront recognition, and what is the renewal rate on data deals signed in 2021–2024?
- Walk through the SB TEMPUS economics: how much revenue/TCV has Tempus recognized from the JV to date, and how do you respond to the claim that the $95M TCV simply mirrored your own $95M contribution?
- Pathos AI is a counterparty in the AstraZeneca foundation-model deal and is affiliated with you personally — what are the related-party terms, who negotiated them on Tempus's behalf, and how is arm's-length pricing assured?
- Operating cash flow worsened to $(218)M in FY2025 even as adjusted EBITDA improved — when does operating cash flow turn positive (not adjusted EBITDA), and what's the bridge?
- Stock-based comp re-accelerated to $52.7M in Q1-2026. What is the multi-year SBC and dilution trajectory, and when does GAAP EPS turn positive?
- Accounts receivable grew faster than revenue and you took $32M in customer equity as payment — what's driving DSO, and what's the policy on accepting illiquid equity for services?
- The Guardant patent suit targets xF/xM MRD — your fastest-growing line. What is your freedom-to-operate position, and what's the downside if an injunction issues?
- What is the genuinely AI-native revenue today (models/software priced as AI, not sequencing), and what's the 3-year plan to make it material?
- Ambry added ~$363M of revenue and $825M of goodwill+intangibles — what return hurdle did you underwrite, and what would trigger an impairment?
- How exposed is Diagnostics to specific CMS rate decisions over the next 24 months, and what's the sensitivity per 5% reimbursement change?
- The FY2026 guide implies ~+25% growth — how much is organic vs. Ambry-in-base, and what's the underlying oncology volume growth?
- With 30-vote super-voting control, what concrete governance commitments will you make to minority holders on related-party deals?
- What does the path to take-out look like given Exact/Abbott — are you building to sell, and does the control structure even permit a sale?
- If the Data line is the real moat, why not segment-report it fully (gross margin, deferred revenue, RPO) so the market can value it as the software business you say it is?