Subscribing to our newsletter will help you navigate and manage your finances with ease through valuable financial tips.
AI & Machine Learning
Fractional CFO for AI & ML Companies
Inference costs, usage-based pricing, and margin compression require a CFO who models compute-driven COGS, not just SaaS metrics.
Trusted by 50+ SaaS businesses, marketplaces, and platforms:
Founder’s pains
Your gross margin is not a SaaS number anymore
Traditional SaaS scales with near-zero marginal cost per user, supporting 80-90% gross margins. AI introduces real per-query compute costs, GPU, memory, energy, that structurally compress margins into a lower band. Your finance model needs to treat inference as a first-class COGS line, not bury it inside generic hosting.
Trigger moments hit hard: AI-savvy investors probe inference costs and contribution margin per model during diligence. Boards question whether your margin trajectory is improving or declining. Pricing model decisions, seat-based, usage-based, hybrid, directly determine whether growth compounds margin or destroys it.
How much inference cost is buried in your hosting line?
AI breaks the near-zero marginal cost assumption that traditional SaaS unit economics depend on. Inference costs, GPU usage, API calls, model routing, memory, scale with every query, yet most AI/ML companies lump them into a generic hosting or infrastructure line item, making it impossible to see true contribution margin per customer or per model.
Are inference whales draining your usage-based revenue?
A small number of power users can generate disproportionate compute costs relative to what they pay: customers who run complex agentic workflows or high-volume queries while on a flat subscription plan. Without per-customer cost tracking and usage caps built into your pricing model, these inference whales quietly consume margin across your entire customer base.
Can your board pack answer AI-specific margin questions?
Investors in AI companies ask questions that traditional SaaS board decks cannot answer: what is your inference efficiency ratio, how does contribution margin differ across models, what is the margin trajectory as usage scales. A board pack that reports historical revenue without modeling compute-driven variable costs leaves the most important strategic questions unaddressed.
Does your runway model capture non-linear compute costs?
Traditional burn models treat infrastructure as a relatively fixed cost that grows modestly with headcount. AI/ML compute costs scale non-linearly with usage growth: agentic workflows can consume dramatically more tokens per task, and competitive pressure forces adoption of more expensive frontier models. A runway forecast that ignores inference-driven variable costs overstates how long your cash actually lasts.
WHY FISCALLION
Decision-grade CFO support built for AI economics
01
Financial Planning and Analysis
A driver-based financial model that accounts for inference costs as a first-class COGS line, usage-based revenue patterns, and compute-driven variable costs. When you ask what happens to runway if you add a frontier model, the model answers with cash impact, margin shift, and the trade-off in hiring or growth spend, not a vague narrative.
02
Fundraising & Investor Support
AI-savvy investor preparation with unit economics defense, inference efficiency narrative, and contribution margin breakdown by model. Every client works directly with Aleksandar at the CFO layer: senior-partner ownership on the board pack and investor deliverable, not a junior analyst running templates.
03
Actionable KPI Dashboards
Dashboards built for AI-specific metrics: inference cost as a percentage of revenue, gross margin by customer cohort, contribution margin by model, usage-adjusted net revenue retention. Each metric connects to a pricing, growth, or capital allocation decision you need to make in the next 30 days, not a vanity chart.
04
Growth Focus CFO Advisory
Senior CFO judgment on pricing architecture, margin recovery strategy, and capital allocation. Aleksandar brings FP&A leadership experience gained through the growth of a SaaS company to €100M ARR, transferable scale judgment applied to AI/ML companies at $5-100M ARR that need CFO-level decision support without a full-time finance org.
FINANCIAL MODELING
AI/ML Financial Infrastructure for Decision-Grade Unit Economics
Inference Cost Model & Efficiency Ratio
Separates inference costs from generic hosting into a dedicated COGS line. Tracks the inference efficiency ratio, AI-related revenue divided by AI-related inference cost, by model and customer cohort, so you can identify which customers and use cases generate margin and which destroy it.
Usage-Based Revenue & Pricing Architecture
Models revenue under seat-based, usage-based, and hybrid pricing structures. Flags inference whales, customers whose compute costs exceed their subscription revenue, and tests pricing changes against margin impact, churn risk, and competitive positioning before you commit.
Gross Margin by Customer Cohort & Model
Reports gross margin at the customer-cohort and model level rather than as a single blended SaaS number. Shows how margin shifts as customers scale usage, adopt more expensive models, or move to agentic workflows, giving you the data to defend your margin trajectory to any AI-specialist investor.
Runway Forecast with Compute-Driven Costs
Extends the runway model beyond fixed burn to include inference costs that scale non-linearly with usage growth. Scenario layers model what happens when token consumption per task rises, when frontier model adoption increases, and when pricing changes shift the cost-revenue relationship.
Board Pack for AI-Savvy Investors
A board deck that frames AI-specific decisions, model selection, pricing architecture, margin recovery, as explicit trade-offs with cash impact. Replaces historical reporting with forward-looking scenarios that answer the question boards actually ask: what do we do next, and what does it cost.
Fundraising Diligence & Unit Economics Defense
Prepares the data room with AI-specific unit economics: inference efficiency, contribution margin by model, usage-adjusted NRR, and margin trajectory. Aleksandar personally owns the investor narrative and diligence response, ensuring consistency between the model, the deck, and the verbal defense.
other sectors
We know tech inside & out.
faqs
Your Questions, Answered
Heading
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Heading
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Heading
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Heading
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Heading
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.
Have more questions? Send us an email
