The Equation Series
Menu

Topic

The business of artificial intelligence

9 articles drawn from The Intelligence Equation, its 24 industry studies and 100 companies profiles.

Explainer

AI Inference Costs Explained: Why Tokens Get Cheaper, Bills Don't

What AI inference costs, why the price per token falls about 10x a year, why reasoning and agents make bills rise anyway, and what it does to gross margins.

9 min read

Guide

AI Wrapper Startups: How to Tell If One Has a Moat

How to evaluate whether an AI wrapper startup has a moat: the thin vs thick test, platform risk from model suppliers, and lessons from Jasper, Windsurf and Cursor.

8 min read

Explainer

CoreWeave's Business Model Explained: How a Neocloud Makes Money

How CoreWeave makes money renting Nvidia GPUs on multi-year contracts, why 60% EBITDA margins still produce losses, and the three clocks behind neocloud risk.

8 min read

Guide

How to Price an AI Product: Seats, Usage or Outcomes

A practical guide to AI pricing: how to choose between seats, usage, credits and outcome pricing, test margins at heavy usage, and learn from Cursor and Sierra.

8 min read

Comparison

Open vs Closed AI Models: How Each Business Model Makes Money

Open-weight vs closed AI models compared: who makes money from each, what licences allow, when self-hosting pays, and why labs open when behind and close when ahead.

8 min read

Case study

What Happened to Inflection AI? The $4 Billion Lab That Microsoft Hired

Inflection AI raised $1.5 billion and built Pi, then Microsoft hired its founders and most staff in 2024. Here is what went wrong and what the company is now.

7 min read

Case study

Why Nvidia Dominates AI Chips: CUDA, Systems and Scarcity

Nvidia earns about 75% gross margins on AI hardware. Here is why: a 2006 software bet, a full-system product, control of supply, and what threatens it.

9 min read

Case study

Why the Humane Ai Pin Failed: $230 Million and 10,000 Units

Humane raised about $230 million, sold about 10,000 Ai Pins and was sold to HP for $116 million in 2025. Here is why the AI device failed, with the numbers.

7 min read