Tempus AI (TEM): Comprehensive Equity Analysis

---

1. Business Model Summary

Tempus operates a two-segment flywheel where the lower-margin diagnostics business generates the proprietary data that powers the higher-margin licensing business—a structure that makes it fundamentally different from either a pure diagnostics company or a pure data/AI company.

Genomics/Diagnostics (75% of FY2025 revenue, $955M): Revenue is volume-driven, billed to physicians and payors upon delivery of next-generation sequencing results. The core products are xT CDx (648-gene solid tumor panel, FDA-approved, $4,500 ADLT rate), xF (liquid biopsy, $3,288 CMS rate, FDA submission pending), and Ambry hereditary testing (acquired early 2025). Oncology ASP was $1,640 in Q4 2025, with management targeting >$2,200 as FDA-approved tests migrate from LDT pricing (~$2,900) and new assays gain coverage (Report 1). GAAP gross margins expanded from 24% in 2022 to 60% in FY2025 as lab automation and volume leverage kicked in (Report 3). Organic oncology volume grew 26% in FY2025, with 340,500 tests in Q4 alone (Report 1).

Data & Services/AI (25% of FY2025 revenue, $316M): This segment monetizes the de-identified multimodal dataset generated by Genomics—over 8 million records spanning clinical notes, genomic sequences, imaging, and transcriptomics (Report 2). Revenue comes primarily through multi-year Insights licensing subscriptions to biopharma (e.g., AstraZeneca $200M/3 years, GSK $70M/3 years), with 126% net revenue retention signaling strong expansion dynamics (Report 1). GAAP gross margins are 72-74%, near software-like economics, because the data is already generated as a byproduct of the testing business (Report 3). Total contract value exceeded $1.1 billion at year-end 2025, up from $940M a year earlier (Report 7).

The critical unit economics insight: Every test Tempus runs simultaneously generates revenue (~$1,640 in oncology) and adds to a proprietary dataset that can be re-licensed at near-zero marginal cost. This creates a structural advantage over pure diagnostics companies (which cap out at ~50% gross margins without a data flywheel) and pure data companies (which must acquire or license data expensively). Report 1 estimates Tempus can achieve 65%+ blended margins at scale versus a 50% cap for pure-play diagnostics. The Insights sub-segment within Data, growing 38% in FY2025 with ~75% margins, is the economic crown jewel (Report 1).

Emerging products include TIME (AI-powered clinical trial matching, $9.2M Q3 revenue), Paige Predict (AI pathology for 123 biomarkers from H&E slides, launched January 2026), and MRD testing (4,700 tests in Q4, +56% QoQ), though these remain early-stage contributors (Reports 1, 2).

---

2. Competitive Positioning and Data Moat Assessment

Where the moat is strongest: Tempus's core defensibility lies in the linkage between its testing volume and its data asset. With 450+ petabytes of multimodal data—genomics, clinical EHR notes (1.6B+ pages), imaging (8M+ records), and transcriptomics (350K+ profiles)—connected across 5,000+ healthcare sites and 55% of U.S. oncologists, Tempus has built what Report 2 describes as a "living dataset" updated in near-real-time, unlike static snapshots from competitors. This data powers AI algorithms (60+ deployed) like the Immune Profile Score, which outperforms PD-L1 alone by integrating histopathology and genomics (Report 2). The 126% NRR confirms pharma clients are expanding their usage of this data, not just renewing (Report 1).

Where rivals hold structural advantages:

Dimension	Tempus Edge	Rival Edge
Dataset breadth	8.5M multimodal records, 90x TCGA (Report 2)	Flatiron: 5M EHR records, deeper clinical curation (Report 2)
Reimbursement maturity	xT CDx approved; xF/xR pending	Foundation: 100 CDx indications; Guardant: pan-tumor NCD (Report 2)
Workflow integration	Tempus One EHR AI, 5K sites	Flatiron OncoEMR embedded in oncology practices (Report 2)
Pharma deal volume	70+ deals, $1.1B TCV	Roche: $4.3B in acquired assets (FMI+Flatiron), primarily internal use (Report 8)

The Roche threat is real but structurally constrained. Report 8 raises the most pointed bear argument: Roche owns both Foundation Medicine (genomics) and Flatiron Health (EHR data), which together could mirror Tempus's multimodal approach. However, no post-September 2025 evidence shows Roche has integrated these assets into a unified external-facing AI platform (Report 8). Roche's internal use of these datasets limits external monetization visibility. Notably, Tempus integrates with Flatiron's OncoEMR for ordering and results—which paradoxically embeds Tempus deeper into the Roche workflow rather than being displaced by it (Report 8). The risk materializes if Roche launches an open-platform equivalent by 2027 (Report 8).

Illumina's decentralization play introduces a different competitive vector. With CMS reimbursement for TruSight Oncology Comprehensive at $2,990 effective January 2026, Illumina enables community labs to run comprehensive genomic profiling in-house, potentially bypassing centralized send-out models like Tempus (Report 2). This structural shift could erode testing volumes at the margins, though it does not replicate the data flywheel.

Guardant leads in liquid biopsy CDx breadth and now competes directly for pharma partnerships (multi-year Merck deal, January 2026), but its $210M biopharma revenue trails Tempus's $316M data segment, and it lacks Tempus's multimodal clinical/imaging linkage (Report 2).

---

3. Financial Trajectory and Key Inflection Points

Revenue acceleration is real but normalizing. Revenue compounded at ~59% CAGR from 2022-2025, with FY2025 hitting $1.3B (+83% YoY). However, the Ambry acquisition contributed materially—organic growth ex-Ambry was 33.5% in Q4 (Report 3). The 2026 guide of $1.59B (+25%) reflects tougher comps and is broadly in line with consensus (Report 3).

Margin expansion is the core financial story:

Year	Total Gross Margin (GAAP)	Diagnostics	Data/Services
2022	40.6%	24.1%	67.2%
2023	53.8%	47.9%	66.6%
2024	55.0%	46.1%	71.5%
2025	62.7%	59.6%	72.3%

(Source: Report 3)

Diagnostics margins nearly tripled from 2022 to 2025—a function of lab automation, volume leverage, and the beginning of ASP uplift from FDA approvals. Data margins are stable at 72-74%, with near-zero marginal cost (Report 3).

The EBITDA inflection arrived in Q3-Q4 2025. Adjusted EBITDA improved $97M year-over-year to -$7.4M for FY2025, with Q4 posting +$12.9M—the first positive quarter (Report 3). The 2026 guide of $65M adjusted EBITDA implies ~4% margin. Report 3 notes the mechanism: roughly two-thirds of incremental gross profit is reinvested, with one-third flowing to EBITDA.

Conditions required for sustained operating leverage:
- Oncology ASP must continue rising toward $2,200+ via xT CDx migration (currently ~30% of volume) and xF FDA approval (Report 1)
- Data segment must grow faster than diagnostics to shift mix toward 35-40% of total, given its 72%+ margins (Report 3)
- MRD reimbursement must arrive (MolDX review ongoing) to unlock volume beyond pilot-scale (Report 6)

Cash burn is concerning but manageable near-term. Operating cash burn was $218M in FY2025 with $760M in cash and marketable securities, implying ~2.8 years of runway at current rates (Report 8). However, this excludes potential ATM usage ($500M remaining on shelf) and a $343M ESOP shelf filed February 2026 (Report 8). Debt stands at ~$1.2B versus $491M equity, yielding a 252% debt/equity ratio (Report 8). Term loans carry covenants tied to minimum revenue ($594M for 2025, met), and convertible notes of $728M net carry conversion overhang of ~9.3M shares (Report 8).

---

4. Valuation Perspective

Current pricing: At ~$52/share (market cap $9.3B, EV $9.9B), Tempus trades at 7.2x trailing P/S and ~6.2x forward EV/revenue on $1.59B 2026 guidance (Report 4). The implied EV/Gross Profit is ~12.4x on $798M TTM gross profit (Report 4).

Peer context:

Company	EV/Revenue (TTM)	Revenue Growth	Gross Margin
Tempus AI	7.8x	83%	63%
Guardant Health	12.7x	~33%	~64%
Veeva Systems	8.3x	~15%	~75%
Schrödinger	2.6x	~23%	~60-70%
Recursion	~15x	~speculative	N/A

(Source: Report 4)

Tempus trades at a discount to Guardant despite triple the revenue growth, likely reflecting Guardant's more mature reimbursement position and Tempus's governance/dilution overhang (Report 4).

What's embedded in current price: The market is pricing ~25% revenue CAGR to 2027 at ~5x forward P/S with 4% EBITDA margins—a discount to comparable high-growth healthcare IT names (Report 4).

Bull case ($78-$100+): Data segment accelerates to 40%+ growth, mix shifts toward higher-margin Insights, EBITDA exceeds $100M, and the stock re-rates to 8-10x forward revenue—aligning with Veeva-like multiples. Analyst targets range to $100 (Mizuho) (Report 4).

Bear case ($30-$40): Growth decelerates to 15-20% on reimbursement stalls or pharma budget cuts, EBITDA remains negative, and the multiple compresses to 4x forward revenue, consistent with Schrödinger/10x Genomics levels. Spruce Point's short thesis implies 50-60% downside (Report 8). Rockflow models a $20-35 base case (Report 4).

The non-obvious valuation tension: The 152x implied EV/EBITDA on 2026 estimates means any miss on the profitability inflection could trigger a severe re-rating, while a beat could compress the gap to profitability-based peers rapidly (Report 4).

---

5. Management and Governance Assessment

The Lefkofsky pattern demands scrutiny. CEO Eric Lefkofsky has founded four companies that went public (InnerWorkings, Echo Global Logistics, Groupon, Tempus). The public track record is mixed at best: Groupon declined 95% from IPO; InnerWorkings was sold at 66% below its IPO price after financial restatements; Echo was taken private at a modest multiple (Report 5). Spruce Point's short thesis explicitly invokes this pattern—"promote, cash out, leave shareholders losses" (Report 5).

Governance red flags are concentrated, not dispersed:
- Dual-class control: Lefkofsky holds 100% of Class B shares with 15-30x voting power, giving him ~60% voting control despite ~24% economic ownership (Report 5). The company qualifies as a "controlled company" under Nasdaq rules (Report 5).
- Related-party entanglement: Pathos AI (co-founded by COO Fukushima, linked to Lefkofsky) generated $41.7M in revenue from Tempus over nine months in 2025. The $200M AstraZeneca deal is structured via the Pathos entity (Report 8). Aircraft usage ($0.8M), legal via brother's firms, and former InnerWorkings CEO Eric Belcher serving as audit chair create a dense web of affiliated relationships (Report 5).
- Insider selling: 596K shares (~$37.5M) sold in the last three months by insiders, including 166K by Lefkofsky in January 2026, while COO Fukushima sold 120K shares in December 2024 (Reports 5, 8).

Mitigants: Lefkofsky retains 24% economic ownership—meaningful skin in the game. No personal legal findings or SEC enforcement actions exist (Report 5). The executive team includes domain hires like Chief Admin/Legal Erik Phelps (ex-Epic Systems GC), which brings healthcare regulatory expertise (Report 5). The company has not invoked the "controlled company" exemption from Nasdaq governance requirements (Report 5).

CFO James Rogers served at Groupon from 2011-2017—the period spanning its IPO, restatements, and CEO turnover. Whether this reflects deep operational experience or institutional proximity to past governance failures depends on one's priors (Report 5).

---

6. Integrated Risk Register

Listed in order of estimated consequence to the investment thesis:

Risk	Likelihood	Impact	Monitoring Signal
Reimbursement stall on xF/MRD — MolDX denial or delays cap diagnostics ASP below $2,200 target	Medium	High — 75% of revenue is diagnostics; ASP trajectory is the single largest margin lever (Reports 1, 6)	Watch MolDX LCD docket quarterly; xF PMA timeline; MRD coverage decisions H1-H2 2026
GIPA/privacy litigation — Nash v. Tempus class action (Feb 2026) alleges GIPA violations from Ambry data transfer; adverse ruling could restrict data licensing model	High (suit active)	High — Data segment is the moat's monetization engine; precedent could constrain future acquisitions' data integration (Report 6)	Track Illinois N.D. proceedings; monitor multi-state class expansion risk
Pharma budget concentration — Top pharma clients (AZ, GSK, Pfizer) comprise outsized share of $1.1B TCV; 49% of TCV due in next 12 months	Medium	High — 20-30% of data revenue at risk if 2-3 majors cut or renegotiate (Report 8)	Monitor pharma R&D budget announcements, particularly post-patent-cliff restructurings (2028 wave)
Roche platform integration — Unified FMI + Flatiron AI platform could erode Tempus's external pharma licensing position	Low-Medium	High — Would challenge the data moat's uniqueness and pricing power (Report 8)	Watch for Roche external platform announcements; FMI/Flatiron partnership disclosures
Dilution and capital structure — $343M ESOP shelf, $500M ATM remainder, $728M convertible notes, $297M unrecognized SBC	High	Medium — Ongoing EPS dilution; conversion overhang ~9.3M shares; GAAP losses persist through 2026+ (Report 8)	Track share count quarterly; SBC as % of revenue; conversion trigger proximity
Illumina decentralization — CMS-reimbursed TruSight Oncology ($2,990) enables community labs to run CGP in-house	Medium	Medium — Could erode centralized testing volume without displacing the data moat itself (Report 2)	Track community lab adoption rates; Tempus volume growth deceleration signals

---

7. Most Differentiated Insights

1. The Genomics segment is actually the wrong place to look for value—Insights is becoming a software business hiding inside a lab company. While 75% of revenue comes from diagnostics, Insights licensing (the sub-segment within Data) is growing at 38-69% with 75%+ margins, 126% NRR, and $1.1B in contracted backlog (Report 1). The marginal cost of each additional data license approaches zero because the data is already generated by the testing business. Report 1 estimates this dynamic can yield 65%+ blended margins at scale. Most investors compare Tempus to diagnostics peers (Guardant, Foundation Medicine) when the economic engine increasingly resembles an enterprise data platform.

2. The GIPA litigation is potentially more consequential than any reimbursement risk. The Nash v. Tempus class action (February 2026) strikes at the legal foundation of the data flywheel—whether genomic data acquired via the Ambry acquisition can be de-identified and licensed to pharma without individual consent under Illinois genetic privacy law (Report 6). If adverse, this wouldn't just create a one-time liability; it could structurally constrain Tempus's ability to integrate data from future acquisitions, the very mechanism that expanded the dataset from ~4M to 9M+ records. Report 6 notes genomic re-identification is possible in 99% of cases via relatives, making the de-identification defense precarious. Peers like Foundation Medicine have avoided similar suits by using research-only covenants and firewalled consents (Report 6).

3. The ASP migration story has a specific, trackable catalyst with asymmetric impact. Only ~30% of xT volume has migrated from the LDT version ($2,900) to the FDA-approved CDx version ($4,500) as of Q3 2025 (Report 1). Full migration, expected by end-2026, represents a >$500 ASP uplift on a base of ~340,000 quarterly tests—pure margin expansion with no incremental volume required (Report 1). Combine this with xF liquid biopsy FDA approval (submitted Q4 2025, adding ~$230 ASP potential) and expanding payer coverage, and the path to >$2,200 ASP has identifiable milestones rather than vague aspirations (Report 1). This is the single highest-leverage financial variable in the near term and is under-discussed relative to top-line growth.

4. Lefkofsky's related-party architecture creates a non-trivial agency risk that doesn't show up in standard financial screening. Pathos AI—co-founded by Tempus COO Fukushima and linked to Lefkofsky—generated $41.7M in revenue from Tempus over nine months and is the entity through which the marquee $200M AstraZeneca foundation model deal flows (Report 8). The AstraZeneca/Pathos deal is one of the largest items in the $1.1B TCV that anchors the bull thesis. Spruce Point's short report and subsequent class action lawsuits specifically allege overstatement or round-tripping involving these arrangements (Report 8). Even if ultimately benign, the opacity of related-party structures creates a governance discount that most financial models don't capture.

5. The RWD market's growth rate may actually understate Tempus's opportunity because the company is creating demand that didn't previously exist. Report 7 sizes the global RWD market at $2.3B in 2026, growing to $4.2B by 2030. But Tempus's multimodal foundation model deals (e.g., AstraZeneca/Pathos for the "largest oncology foundation model") represent a category of spending—AI model training on linked clinical/molecular/imaging data—that barely existed two years ago. Report 7 notes that the related AI-powered RWE solutions market is $5.4B in 2025, growing at 14.8% CAGR. Tempus's $1.1B TCV against a $2.3B market suggests either significant market share capture or that the TAM definitions are trailing the actual addressable spend for multimodal AI-enabled data services.