Moluscca MOLUSCCA Quantitative Risk Assessment
Enterprise Risk Intelligence Report · Confidential Draft

Aerthos: what if we
built autonomous vertical
farming
at scale?

A full pre-launch risk assessment of a synthetic climate-tech venture, evaluated as a professional risk-intelligence firm would evaluate it: from first-principles business model through Monte Carlo survival analysis, tail modeling, and a weighted enterprise risk score.

SubjectAerthos Systems (synthetic)
SectorAutonomous CEA / AgTech
Horizon60 months
Overall Risk62.1 / 100 · High
Prepared byMoluscca Risk Intelligence
⚠ All data in this report is synthetic, generated by Monte Carlo and stochastic simulation for methodology demonstration. It is not real-world data and describes no real company. Figures are internally consistent with the simulated model only.
01 · Executive Summary

The verdict in one screen

Aerthos is a moderate-to-high risk venture whose survival hinges almost entirely on liquidity and energy cost, not on demand. The technology works in simulation; the balance sheet is the fragile part.

Overall Risk Score
62.1
High · weighted across 10 dimensions
P(survival · 5yr)
65%
100k Monte Carlo paths
P(insolvency)
35%
cash < 0 before month 60
VaR (95%)
$38M
capital-at-risk vs. $14M raised
CVaR (95%)
$46M
expected loss in worst 5%
Projected ARR
$13.6M
annualized, month-60 run-rate
P(next raise)
98%
at least one Series A/B success
Blended churn
6.9%
across 10,000 offtake contracts
◆ AI-generated insight

The defining tension is frequency vs. severity. Aerthos faces many small, survivable operational losses (crop failures, downtime) but its ruin is driven by rare, severe liquidity shocks. A 35% insolvency probability against a 98% financing-success rate means the business is fundable but under-capitalized for its own volatility: the single highest-leverage intervention is extending runway, not growing revenue.

02 · The Business, From First Principles

What Aerthos is

Aerthos designs and operates shipping-container-scale autonomous vertical farms, sold to grocery, foodservice, and QSR buyers as fixed-price offtake contracts for leafy greens grown within 50km of the point of sale.

Revenue model

Two streams: (1) recurring offtake contracts priced per kilogram, blended at $4.2/kg wholesale; (2) a hardware-plus-software platform fee per deployed module. Revenue scales with modules deployed on a logistic adoption curve toward a 42-farm network by month 60.

Cost structure

Energy is the dominant variable cost (vertical farming is electricity-intensive), followed by labor and maintenance. This concentration is why an energy-price tail, modeled below with Extreme Value Theory, is a genuine solvency risk rather than a line-item nuisance.

Customer segments

SegmentContractsAvg MRRChurn
Foodservice2,712$2,02310.0%
Grocery Retail4,222$3,0946.0%
Institutional1,220$1,6906.0%
QSR Chain1,846$10,0275.0%

Simulated ledger of 10,000 synthetic offtake contracts. Foodservice shows the highest churn and is the segment most sensitive to price and service reliability.

03 · Financial Trajectory

Revenue, cost, and the cash line

The base-case path: revenue crosses opex, but the cash trough before that crossover is where most simulated failures occur.

Cash runway
Base-case monthly revenue, opex, and cash position over 60 months. The cash line dipping toward zero around the mid-period trough is the venture's structural vulnerability.
04 · Monte Carlo Survival Analysis

100,000 possible futures

Each path randomizes demand, cost, financing success, and rare shocks. The distribution of outcomes, not any single forecast, is the answer.

MC cash
Final cash position across 100k simulations. Mass left of the red line is insolvency. P(survival) = 65%.
MC loss
Capital-loss distribution with VaR95 = $37.9M and CVaR95 = $46.2M marked.

Why Monte Carlo

Selected because the venture's outcome is a path-dependent compound of many correlated uncertainties with no closed-form solution.
P(survival) = (1/N) Σ 1[ cashₜ ≥ 0 ∀ t ∈ 1..60 ], N = 100,000
Cash evolves as cashₜ = cashₜ₋₁ + Revₜ·δ − Costₜ·γ, with δ~N(1,0.28), γ~N(1,0.16), and a 1%/month severe-shock multiplier. Result: 65% survival, expected final cash $6.6M.
05 · Tail Risk, VaR, CVaR & Extreme Value Theory

The edge, where ventures actually die

Averages are comfortable; tails are fatal. Two tail engines run here: a Loss Distribution Approach for operational losses, and a Generalized Pareto fit for the energy-price extreme.

Op loss
Annual aggregate operational loss (Poisson frequency × Lognormal severity). P99 = $4.73M, P99.9 = $6.76M.

Extreme Value Theory · Energy price tail

Peaks-over-threshold: fit a Generalized Pareto to exceedances above the 90th percentile of the energy-cost shock.
P(X−u > y | X > u) = (1 + ξy/σ)^(−1/ξ)
Fitted shape ξ = 0.149 (ξ>0 confirms a heavy tail), scale σ = 2.57. The modeled 1-in-100 energy shock reaches 22.25× the baseline unit cost, a level that alone can flip several stress scenarios into insolvency.

Loss Distribution Approach

S = Σᵢ₌₁ᴺ Xᵢ, N ~ Poisson(λ=6), Xᵢ ~ Lognormal
Mean annual op-loss $1.62M, tail to $6.76M at P99.9.
06 · Drivers, Correlation & Anomaly Detection

What moves with what

Correlation matrix
Correlation matrix of the eight core risk drivers. Energy and opex move together; cyber events cluster with downtime, which matters for scenario design.

Logistic Regression · Churn / default proxy

Chosen for interpretability: every coefficient reads as a log-odds effect a regulator or underwriter can follow.
logit(p) = β₀ + β₁·MRR + β₂·tenure + β₃·segment
Tenure lowers churn odds; the Foodservice segment raises them, quantifying the earlier churn table. Coefficients: MRR -0.0000, tenure -0.0170.

Isolation Forest · Operational anomaly

Unsupervised; flags abnormal months without labeled failures.
Flagged anomalous months: M40, M50, M52, each coinciding with a cost or cyber spike, useful as an early-warning trigger.
07 · Enterprise Risk Register & Scoring

Ten dimensions, one weighted score

Each risk scored 0–100 from likelihood, severity, velocity, and detectability, then weighted into a single enterprise figure of 62.1 (High).

Likelihood →
FIN
ENE
OPE
REG
TEC
CYB
MAR
SUP
CLI
HUM
low↑ Severityhigh
Risk dimensionLikelihoodSeverityExp. lossScoreBand
Financial / Liquidity
0.62
0.88
$10.9M 73 Critical
Energy / Input Cost
0.70
0.72
$10.1M 69 High
Cyber / OT Security
0.45
0.78
$7.0M 66 High
Market / Demand
0.58
0.62
$7.2M 60 High
Operational / Yield
0.55
0.65
$7.2M 58 High
Supply Chain
0.60
0.50
$6.0M 57 High
Technology / Automation
0.48
0.60
$5.8M 56 High
Regulatory / Food Safety
0.40
0.80
$6.4M 56 High
Climate / External
0.35
0.70
$4.9M 53 Moderate
Human Capital
0.50
0.45
$4.5M 48 Moderate

Scoring methodology

scoreᵢ = 100·(0.42·Lᵢ + 0.38·Sᵢ + 0.12·Vᵢ + 0.08·(1−Dᵢ))^0.92
Overall = Σ (scoreᵢ · wᵢ) / Σ wᵢ
L=likelihood, S=severity, V=velocity (speed of onset), D=detectability (higher=easier to catch, so it reduces risk). The 0.92 exponent gently compresses mid-range scores. Weights wᵢ reflect each dimension's contribution to solvency, with Financial/Liquidity weighted highest at 0.17.
08 · Composite Indices

Readiness at a glance

62 RISK / 100
Overall Risk62.1
Revenue volatility6.6%
Resilience index56.5
Stability (survival)65.4
Insurance readiness57.8
Investment readiness62.3
09 · Stress Testing & Scenario Analysis

Eight ways the world gets harder

Each scenario re-runs the full model under altered demand and cost assumptions. Survival probability is a fresh 20,000-path Monte Carlo per scenario.

ScenarioFinal cashRunwayP(survival)Outcome
Base Case $-18.3M 42 mo
46%
Insolvent
Funding Winter $-27.1M 38 mo
26%
Insolvent
Energy +40% $-30.6M 37 mo
20%
Insolvent
Interest Rates +5% $-25.2M 39 mo
30%
Insolvent
Major Cyber Attack $-22.8M 39 mo
32%
Insolvent
Demand Crash -35% $-30.9M 36 mo
20%
Insolvent
Regulatory Halt (3mo) $-36.0M 35 mo
13%
Insolvent
Global Recession $-32.5M 36 mo
17%
Insolvent
◆ Reading the stress table

Demand shocks are survivable; cost and liquidity shocks are not. A 35% demand crash still leaves positive runway, but a major cyber attack or an energy spike compounding a funding winter is what pushes survival probability below 50%. This is the actionable core: Aerthos should insure and hedge cost-side tail risk long before it worries about demand.

10 · Executive Recommendations

What we would tell the board

01

Extend runway before scaling

The dominant failure mode is the mid-period cash trough. Raise a bridge or slow deployment so cash never approaches zero; this single move lifts survival probability more than any revenue gain.

02

Hedge and insure energy

With EVT shape ξ=0.15, the energy tail is heavy. A power-purchase agreement or parametric energy-cost cover converts an unbounded tail into a fixed premium.

03

Harden OT / cyber

Cyber has the highest velocity score and clusters with downtime. It is low-frequency but high-severity, exactly the profile insurance is built for. Secure coverage and segment the control network.

04

Defend Foodservice retention

The logistic model localizes churn risk to one segment. Service-level guarantees there protect the revenue line the survival model is most sensitive to.

05

Stagger the raise

Financing success is high (98%) but timing-dependent. Pre-arrange the month-24 round rather than relying on being cash-positive to unlock it.

06

Instrument early-warning

Wire the Isolation-Forest anomaly flags into live monitoring so a cost or cyber spike triggers action in the month it happens, not at quarter close.

◆ Final classification

Aerthos: High risk (62.1/100). Fundable, conditional on liquidity discipline. The venture is not doomed by its market or its technology; it is threatened by its own volatility against a thin capital base. Priced correctly, most of that volatility is insurable or hedgeable. Investment-readiness 62.3, insurance-readiness 57.8.