How to Explain AI Gross Margin, Inference Cost and COGS to Investors

TL;DR

• Separate AI/model costs from other cloud and operating expenses when they materially drive COGS.
• Show gross margin using consistent accounting and explain how usage intensity affects it.
• Model what happens if model prices, context size, traffic mix or customer behavior changes.
• Investors care about the path to durable economics, not just today's inference bill.

Build the AI economics story from the workload up

AI gross margin can be misunderstood when founders report SaaS-style revenue while model and infrastructure costs scale differently from traditional software. A good fundraising explanation shows the actual workload, cost drivers and improvement levers.

Practical workflow

1. Define the workload

Choose a unit such as request, document, minute, workflow or active customer that maps to actual usage.

2. Map variable AI costs

Include model API, GPU/inference, vector/database and other directly usage-linked infrastructure where appropriate.

3. Calculate gross margin consistently

Use the company's accounting policy and explain any non-standard treatment.

4. Show segmentation

Compare high-usage versus low-usage customers or workflows if economics differ materially.

5. Model improvement levers

Explain routing, caching, smaller models, batching, fine-tuning or pricing changes that can improve unit economics.

6. Stress-test the plan

Show sensitivity to model-provider pricing and usage growth.

What to prepare

• COGS definition and accounting treatment.
• Model/provider invoices or usage summaries.
• Unit-cost model.
• Customer usage segmentation.
• Pricing and gross-margin scenario model.
• Financial forecast.

Related SendNow resource: AI startup fundraising workflow.

Common mistakes

• Quoting gross margin without explaining AI-related COGS.
• Assuming model costs will always fall.
• Ignoring customer-level usage variation.
• Presenting optimization ideas as guaranteed future savings.

External reference: NIST AI Risk Management Framework. This article is educational and not legal, regulatory or financial advice.

See the VDR and Microsite workflow

Frequently asked questions

What counts as AI COGS?

That depends on the company's accounting policy and business model; founders should use consistent treatment and qualified accounting advice.

Should model API spend be shown separately?

If it is material to the economics, separating it can make the scaling story easier to understand.

What do investors want to see?

A credible current margin, clear cost drivers and evidence that the business has levers to improve economics as it scales.

Build a cleaner AI investor workflow

Use SendNow Microsites when AI fundraising moves beyond the deck into multi-file diligence.