FIG Analysis: FIG — SaaS Survival V-Score: 2.42 (κ=0.00)

V-Score Card

TICKER:          FIG (Figma, Inc.)
V-SCORE:         2.42
VERDICT:         AT_RISK
κ (conviction):  0.00  →  NO BASKET INCLUSION
Bustamante:      0/3 — no proprietary data, no regulatory mandate, no transaction rail

DIMENSIONS          SCORE   WEIGHTED
─────────────────────────────────────
C  Cognition  ×0.25   3      0.75
E  Infrastr.  ×0.22   2      0.44
U  Breadth    ×0.18   3      0.54
A  Distrib.   ×0.12   3      0.36
M  Gravity    ×0.15   3      0.45
F  Friction  ×−0.06   2     −0.12
─────────────────────────────────────
Raw sum                      2.42

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

V = 2.42 × 1 × 1 = 2.42     AT_RISK
κ = max(0, 2.42 − 3.0) = 0.00

Regime Context

Factor regression over 15 weeks (69 trading days, Dec 24 – Apr 8):

FIG ~ SPY + IGV
─────────────────────────────────
α̂ (annual):    −54.1%   (t = −0.44, p = 0.66)
β_SPY:           0.29
β_IGV:           1.30
σ_idio (ann):   62.5%
%Idio Var:      67.1%    (below 75% target)
R²:              0.33

IR_FIG = α̂ / σ_idio = −0.87

ρ_intra (market-neutralized, 11 software names): 0.50
Rolling 20-day: 0.54 (Feb) → 0.52 (Mar) → 0.44 (Apr)

IR is measuring the regime, not the stock. ρ_intra = 0.50 signals an indiscriminate sector selloff — when software names move as a block, idiosyncratic residuals collapse toward zero and IR becomes unreliable. The α is statistically insignificant (p = 0.66). The −54% annualized "alpha" is sector carnage bleeding through an underpowered regression, not a stock-specific signal.

Period returns confirm the pattern: FIG −45%, TEAM −60%, ASAN −55%, MNDY −54%, ZS −38%, IGV −26%, SPY −4%. This is a sector-wide rout, not a company-specific repricing.

IR does not gate the verdict. V-Score is structural, regime-invariant. V(s) ⊥ r_sector(t). The selloff is indiscriminate but the structural weakness is real — both IR and V point in the same direction for this name.

δ = V − V_market: When ρ_intra → 1, the market applies a uniform discount to the sector. For names with V > 3.0, this creates a structural mispricing opportunity (δ > 0, market overpenalizes durable businesses). For FIG, V = 2.42 < 3.0 — the uniform discount and the structural reality are aligned. No exploitable δ.

Dimension Analysis

C = 3 — Compound Cognition

10+ years accumulated. WebGL rendering engine bypasses the DOM to render directly to GPU — the technical innovation that displaced Sketch and Adobe XD (10-K ln 649-651). CRDT-based multiplayer for real-time conflict resolution across distributed clients. 9 products with superlinear inter-module dependencies: Make → Design (Copy Design, 10-K ln 542-544), Design → Make (Make Kits, 10-K ln 540-542), FigJam → Slides (AI-powered transition, 10-K ln 445-448), Design → Dev Mode → Code (MCP + Code Connect, Q4 transcript ln 36-38). 70% of customers use 3+ products (Q3 transcript ln 65). 22 issued US patents, 237 pending (10-K ln 1099-1102).

The limitation. The AI layer — Figma's fastest-growing capability — is built on commodity models: "We primarily rely on off-the-shelf foundational models" (10-K ln 654-656). No deep domain encoding. SAP has 40 years of business process logic across industries. Figma has canvas rendering, vector math, and constraint solving — engineering challenges, not crystallized domain knowledge.

c_derive timeline: Core canvas 6-12 months. CRDT multiplayer 12-18 months. Full 9-product platform with edge cases: 2-3 years. The threat isn't re-derivation. It's bypass — AI code gen makes the intermediate design step optional, and c_bypass is falling faster than c_derive can protect.

E = 2 — Irreducible Infrastructure

The dominant discriminator, and the structural anchor.

Zero tasks in Figma's domain have c_ℓ = ∞. No regulatory mandate — FedRAMP is a qualification to sell to government, not a mandate to buy from Figma (10-K ln 797-800). No petabyte-scale infrastructure — design files live on standard AWS (10-K ln 7644-7645). Files are exportable (SVG, PNG, PDF). 97% GRR among $10K+ customers (Q4 transcript ln 61) reflects current willingness to stay, not structural inability to leave.

What makes FIG's 10-K remarkable is management's candor. Four erosion mechanisms are named explicitly in the risk factors:

Seat compression. "Organizations may reduce the number of designers, developers, or collaborators required to deliver products, which may lead to them reducing the number of seats purchased or renewed even if output increases" (10-K ln 1883-1885).
Platform consolidation. "Customers may consolidate purchasing decisions around vendors that bundle AI capabilities across broader software suites or purchase from other vendors that provide end-to-end 'design-to-code' or 'prompt-to-product' experiences" (10-K ln 1879-1881).
Billing model shift. "Customers may shift preferences from per-seat subscriptions to credit-based, usage-based, or outcome-based billing models" (10-K ln 1887-1889).
Declining switching costs. "Switching costs may decline as AI tools improve interoperability, automated conversion of files, or automated recreation of design systems across platforms" (10-K ln 1892-1893).

E = 2 is correct today but fragile. The distance between E = 2 and E = 1 is the collaboration friction of migrating team workflows and design systems — currently measured in weeks-to-months. If AI tools automate design system conversion, that timeline compresses to days, and E slides to 1. Gate 1 requires E > 1. If E = 1, V = 0 (COLLAPSED). This is a cliff, not a slope.

U = 3 — Ecosystem Breadth

9-10 products spanning 5-6 departments: Design, Draw, Make, Dev Mode, FigJam, Slides, Buzz, Sites/Payload, Weave (10-K ln 831, 463, 520, 492, 428, 437, 591, 561, 634). Non-designers are 2/3 of monthly active users. 70% multi-product adoption (Q3 transcript ln 65) creates superlinear switching cost — migrating one product is easy; migrating five interconnected products with cross-module data flows is substantially harder.

Concentrated in the product development domain. Not company-wide breadth (contrast ServiceNow spanning IT, HR, CSM, security). Many products launched in 2025, not yet deeply penetrated.

A = 3 — Distribution

Dev Mode MCP server (10-K ln 836-841), Claude Code canvas-to-code roundtrip (Q4 transcript ln 19-22), GitHub MCP Registry partnership (Q4 transcript ln 39), ChatGPT Figma App (10-K ln 876-877), Code Connect for GitHub Copilot (Q4 transcript ln 38). Integrations with Gemini, Notion, Linear, Supabase (Q3 transcript ln 50).

Agents can consult Figma but don't need to route through it. An agent building UI from a text prompt generates code directly — Figma is additive context, not required infrastructure. The MCP roundtrip is a defensive pivot: keeping Figma in the loop as AI code gen threatens to bypass design entirely. It is the right strategy. It is not yet proven at scale.

M = 3 — Ecosystem Gravity

$1.056B FY2025 revenue, +41% YoY (Q4 transcript ln 58). 540K+ paid customers. 13,861 customers > $10K ARR, 1,405 > $100K, 67 > $1M (10-K ln 5694-5698). NRR 136%, improving from 122% in 2023 (10-K ln 5698). Apple provides native design resources through Figma (10-K ln 809-810). Config conference, community chapters worldwide, Figma Ventures ecosystem fund.

No counterparty network effects. Migration is internal to the organization — unlike ADP (payroll connects to banks/government) or Visa (connects merchants to cardholders), Figma doesn't connect you to external parties. Total migration cost: weeks to months, not years.

F = 2 — Ecosystem Friction (penalty)

Browser-based, freemium, self-serve, clean APIs, MCP integration, no consultant dependency (10-K ln 649, 703, 687, 666, 836, 748). Near-zero individual onboarding. This is what killed Sketch and built Figma's dominance — and it is symmetrically what makes Figma replaceable. Low friction to adopt = low friction to leave.

Sensitivity

Scenario                              C  E  U  A  M  F    V     Verdict
─────────────────────────────────────────────────────────────────────────
Base case                             3  2  3  3  3  2   2.42   AT_RISK
Bull: MCP becomes required routing    3  2  3  4  3  2   2.54   AT_RISK
Bull: Make/Sites mature               3  2  4  3  3  2   2.60   AT_RISK
Bull: Governance → regulatory pull    3  3  3  3  3  2   2.64   AT_RISK
Bear: AI kills design seats           3  1  3  3  2  2   2.07   AT_RISK
Extreme bear: E fails gate            3  1  3  2  2  2   0.00   COLLAPSED

Even bull scenarios cannot push FIG above AT_RISK. The path to EMBEDDED (V ≥ 3.0) requires E ≥ 3, which requires regulatory mandate or petabyte-scale proprietary infrastructure — implausible for a design tool.

Thermodynamic Summary

Collaboration friction — the cost of migrating team workflows, design systems, and organizational muscle memory — is the only barrier preventing intelligence from flowing around Figma. The core function (translating ideas into visual interfaces) is directly in the kill zone of AI code generation. The Tool Death Theorem applies: as c_ℓ(design, t) → 0 via v0, Bolt, Lovable, and Cursor, the intermediate step of "design in Figma, then hand off to code" becomes optional for an expanding fraction of software development.

Figma's MCP integrations are the right defensive pivot — making the canvas a roundtrip station in the AI coding loop. But this positions Figma as a useful context layer, not a required routing point. The distinction matters: ServiceNow is a required routing point (agents must file tickets through it). Figma is an optional context source (agents can generate UI without it).

Durable revenue (≈35-40%): Enterprise collaboration/governance layer (design reviews, versioning, permissions, Governance+), design system management as organizational IP, Dev Mode/MCP bridge for human-in-the-loop review of AI-generated code. Survives because coordination ≠ creation.

Exposed revenue (≈60-65%): Pure design seats (individual UI mockups), FigJam (whiteboarding), Slides (presentations), Buzz (marketing collateral), any seat where primary use = "create visual artifacts." AI compresses headcount → fewer seats → revenue declines even if output increases.

Kill cycle: Graduated, not binary. Individual design tasks: t* ≈ 1-2 years (already happening). Team coordination: t* ≈ 3-5 years. Enterprise governance: t* ≈ 5+ years. Figma will erode from the individual/SMB tier upward as the simplest design tasks migrate to AI code gen first.

Conviction Weight

V   = 2.42          AT_RISK
κ   = (V − 3.0)⁺ = 0.00
w   = 0             NO BASKET INCLUSION
IR  = −0.87         (noise — ρ_intra = 0.50, p = 0.66)
δ   = moot          (κ = 0 → no structural edge to exploit)

Basket Verdict

Exclude. V = 2.42 is structurally anchored in AT_RISK territory. The E dimension (irreducible infrastructure) is the binding constraint, and a design tool has no plausible path to E ≥ 3. Even aggressive bull assumptions (MCP becomes routing point, Make/Sites mature, governance creates regulatory pull) only push V to 2.64 — still AT_RISK.

The business metrics are excellent: $1B+ revenue growing 41%, 88% gross margins, 136% NRR, 97% GRR, $1.7B cash, no debt. By every traditional SaaS metric, Figma is a high-quality company. V-Score does not measure current performance. It measures structural resistance to intelligence compression. On that axis, a design tool scoring 0/3 on Bustamante — no proprietary data, no regulatory mandate, no transaction embedding — is AT_RISK by construction.

The per-seat model is the structural vulnerability. Management knows it (10-K ln 1883-1889). The market knows it (−85% from IPO high). V-Score confirms what both are telling you.

Sources: FIG 10-K (filed 2026-02-18), Q4 2025 earnings transcript, Q3 2025 earnings transcript, yfinance market data (2026-04-08). Factor regression: FIG ~ SPY + IGV, T = 15 weeks (69 trading days). Intra-sector correlation computed across 11 software names (FIG, ADBE, TEAM, MNDY, ASAN, GTLB, CRM, NOW, DDOG, NET, ZS), market-neutralized residuals.