Score Card

PTC V-SCORE: 3.10  |  EMBEDDED  |  κ = 0.10

  C = 4   Compound Cognition        (w=0.25)   →  1.00
  E = 3   Irreducible Infrastructure (w=0.22)   →  0.66
  U = 4   Ecosystem Breadth          (w=0.18)   →  0.72
  A = 3   Distribution & Discovery   (w=0.12)   →  0.36
  M = 4   Ecosystem Gravity          (w=0.15)   →  0.60
  F = 4   Ecosystem Friction         (w=−0.06)  → −0.24
                                                  ──────
  V = 1.00 + 0.66 + 0.72 + 0.36 + 0.60 − 0.24 =  3.10

  Gate 1: E = 3 > 1  → PASS
  Gate 2: A = 3 > 1  → PASS

  Verdict: EMBEDDED
  κ = (3.10 − 3.0)⁺ = 0.10
  Basket: KEEP at minimum weight

Regime Context

IR does not gate the verdict. It measures the regime, not the stock.

Over the trailing 15 weeks (Dec 16 – Mar 27, 70 trading days), PTC returned −21.4%. The decomposition:

PTC's entire drawdown is factor. Cumulative idiosyncratic return is zero. The regression:

r_PTC = α̂ + 0.863 × SPY + 0.673 × IGV⊥ + ε

α̂ (ann):      −62.4%   (t = −1.50, p = 0.14 — NOT significant)
σ_idio (ann):   21.5%
%Idio Var:      51.4%   (below 75% target — factor regime)

IR = α̂ / σ_idio = −2.90  (noise, not signal)

The intra-sector correlation diagnostic confirms: ρ_intra = 0.61 (raw), rising to 0.81 in the most recent 20 days. When ρ_intra → 1, all names move as a block, ε_i → 0, and IR measures the regime — not alpha, not the stock, not the thesis.

V(s) ⊥ r_sector(t). The V-Score is computed against structural properties — product vintage, contract lock-in, workflow breadth, agent routing, ecosystem gravity. None of these changed because IGV sold off 27%. The geometric kernel doesn't simplify. The regulatory traceability records don't evaporate. The 3-year contracts don't shorten.

Edge δ = V_PTC − V̄_market. During indiscriminate selloffs, the market applies a uniform discount across the sector. PTC (V = 3.10, EMBEDDED) trades alongside names that score 2.5–2.8 (AT_RISK). The structural gap δ ≈ 0.3–0.6 is the edge the market will price when ρ_intra normalizes and names differentiate. The selloff creates δ; mean-reversion in correlation surfaces it.

Dimension Analysis

C = 4 — Compound Cognition. 40 years of parametric CAD encoding (Creo, ≈1987). 27 years of PLM workflow crystallization (Windchill, ≈1998). The geometric kernel — parametric constraint solver, B-rep modeling, tolerance analysis, FEA simulation — represents decades of mathematical encoding that is not trivially re-derivable. "3D CAD technology enables the digital design, testing, and modification of product models. With its design simulation, additive manufacturing, and generative design innovations" [10-K FY25, L215-220]. Inter-module dependencies create superlinear re-derivation cost: Creo → Windchill → Codebeamer → ServiceMax form a hub-and-spoke architecture where Windchill is the gravitational center. "Our core products, CAD, PLM, ALM, and SLM, are the systems of record across the life cycle" [CEO Barua, Q1 FY26]. Compound re-derivation: 4–6 years for the integrated system. Not 5 because Siemens (Teamcenter/NX), Dassault (3DEXPERIENCE/CATIA), and Autodesk (Fusion) have comparable depth — "Autodesk, Dassault Systèmes SA, and Siemens AG" [10-K, L258-259].

Adversarial check (C): Frontier AI compresses Layer 1 (mathematical foundations — NURBS, constraint solvers). Open CASCADE exists. But Layer 2 (engineering heuristics — millions of edge cases accumulated over decades of production use) requires time in the field, not compute. Layer 3 (workflow crystallization — how a change order flows from engineering to manufacturing at Boeing) requires customers, not code generation. The compound re-derivation across all three layers holds. PTC's customers design jet engines and pacemakers — a 95%-correct constraint solver is 0% useful when tolerance stack-ups fail on a turbine blade. C = 4 survives.

E = 3 — Irreducible Infrastructure. "95% of our 2025 revenue is recurring" [10-K, L178]. RPO: $2.94B as of Dec 2025 [10-Q, L413], with ≈55% recognized in the next 12 months. Average contract duration stepped from 2 years to 3 years in FY25 [CFO Talvitie, Q4 FY25]. "Record deferred ARR under contract... triple what we had last Q4" [CEO Barua, Q1 FY26]. Customers "need to maintain subscriptions to continue designing, producing, and servicing their products" [CFO Talvitie, Q4 FY25]. Not 4 because: NRR/GRR undisclosed (companies with >130% NRR trumpet it — the silence is a number), no regulatory mandate (Arena is regulatory-adjacent but FDA doesn't require PTC specifically), infrastructure is replicable (Siemens Teamcenter, Dassault 3DEXPERIENCE), and ServiceMax shows active churn — "not out of the woods" [CEO Barua, Q1 FY26].

Adversarial check (E): Switching cost is real but finite. Estimated 18–36 months and $10–30M+ for a large enterprise migration (data, ERP reintegration, engineer retraining, regulatory revalidation). High, not infinite. The Garrett Motion displacement proves the cost is surmountable in both directions — customers DO switch, which means the moat is wide but not uncrossable. E = 3 holds at the prior.

U = 4 — Ecosystem Breadth. Seven products spanning PLM, ALM, SLM, service parts, QMS, CAD (parametric + cloud-native). Revenue split: PLM $1,741M (63.6%), CAD $998M (36.4%) [10-K, L3527-3529]. Coverage across ≈6 departments (engineering, manufacturing, quality, field service, supply chain, product management) and 5 verticals (industrials, FA&D, electronics, automotive, medical devices). "27 of the top 30 public medical device manufacturers use PTC" [Q1 FY25]. The digital thread — Creo → Windchill → Codebeamer → ServiceMax — is a genuine cross-functional data architecture from concept through end-of-life. "CodeBeamer 3.2 deepens the connection between CodeBeamer and Windchill" [Q1 FY26]. Not 5 because the lifecycle is narrow compared to a full enterprise platform — engineering and manufacturing, not HR, finance, or sales.

A = 3 — Distribution & Discoverability. PTC is building agent infrastructure: "common AI infrastructure across our product portfolio... AI agents to understand and use product data from CAD, PLM, ALM, SLM, and third-party systems" [CEO Barua, Q1 FY26]. AI features shipped in all five product areas (Windchill AI parts rationalization, Codebeamer AI, ServiceMax AI, Onshape AI Advisor, Arena AI). But AI revenue is explicitly "immaterial" [CEO Barua, Q1 FY26]. No evidence agents currently route through PTC by default. The thesis — that agents will need structured product data to operate in manufacturing — is genuine but 2–3 years from materialization. Today, PTC has functional APIs, not default routing.

M = 4 — Ecosystem Gravity. 30,000+ customers globally [10-K, L145]. $2.49B ARR growing +13% YoY (+8% CC) [10-Q, L1114]. Competitive displacement accelerating: "record-setting Q1 of large deal volume and strong competitive displacements" [CEO Barua, Q1 FY26]. Switching costs are multi-dimensional: proprietary CAD data formats (years of models), ERP/MES integrations, engineer retraining (parametric muscle memory), regulatory audit trail revalidation. "PLM tend sticky, mission critical" [Q3 FY24]. Not 5 because no counterparty network effects — a competitor using Creo doesn't make Creo more valuable to you. Single-firm gravity, not multi-party.

F = 4 — Ecosystem Friction (penalty). Consultant-dependent implementations. Professional services at 1% gross margin, $107M and declining 19% YoY by design as PTC shifts to partner-led delivery [10-K, L1359-1361]. SaaS transition early: "the dam has not broken where the entire market is flipping to our plus platform overnight" [CEO Barua, Q1 FY26]. Windchill Plus shows 1.5–2.5× ARR uplift on conversion. Steep learning curve on parametric CAD acknowledged: new UI/UX in progress [Q4 FY25]. At w = −0.06, this costs 0.24 points — real but not determinative.

Evidence Gaps

Sensitivity

One dimension drops by 1 in C or E → AT_RISK. PTC sits on the boundary.

Comparable Universe

PTC slots with Salesforce/Workday: deep enterprise software, sticky customers, no regulatory mandate or transaction embedding. Well above Adobe's creative-tool substitution risk. Well below Synopsys/Cadence's regulatory moats.

Thermodynamic Summary

PTC's survival mechanism is compound cognition (40 years of parametric CAD/PLM encoding) plus ecosystem gravity (30K customers locked by product data, regulatory traceability, and engineer training). Intelligence flows through PTC because re-deriving from first principles is expensive (4–6 years for the integrated system) and the switching cost φ is measured in years and tens of millions of dollars.

The binding constraint is E = 3: no regulatory mandate, no unique data asset, infrastructure replicable by Siemens and Dassault. If an AI-native CAD/PLM platform emerges that can ingest Windchill data with better UX — and PTC's SaaS transition stumbles — the gravitational well leaks. The upside path: agents routing through manufacturing systems of record (A → 4, V → 3.22). The downside path: AI compresses re-derivation cost (C → 3, V → 2.85).

Durable revenue (≈70%): Enterprise Windchill PLM (deep customization, years of product data, regulatory traceability), large-enterprise Creo seats (parametric expertise, simulation dependencies), Codebeamer in regulated automotive/medical.

Exposed revenue (≈30%): SMB CAD (Onshape-class, cloud-native AI-first tools emerging), ServiceMax (churning, competitive with Salesforce FSM/IFS), professional services ($107M, declining 19%/yr, agents accelerate this), Arena SMB PLM.

V = 3.10. EMBEDDED. κ = 0.10. KEEP at minimum basket weight.

The score measures structure, not price. The selloff is regime (ρ_intra = 0.81, factor dominance). When correlation normalizes, δ surfaces as alpha. Until then, κ sizes the position — regime-invariant, structurally grounded, honest about the boundary.