About canar.ai
The canary in the AI coal mine. Tracking the means of intelligence production.
What is this?
canar.ai maps every publicly traded company in the AI supply chain across a 9-layer stack — from the rare earth mines that produce raw materials to the applications deploying AI in the real world. Think of it as CoinGecko, but for AI stocks.
Unlike generic stock screeners, canar.ai classifies companies by their role in the AI ecosystem and estimates how much of each company's revenue comes from AI. This helps investors distinguish real AI plays from companies that are simply "AI-washing" their narrative.
AI Fair Value
AI Fair Value is our ranking metric. It answers: "How much of this company's market cap is attributable to AI?"
AI Revenue % is estimated using a Bayesian hierarchical model that combines SEC filing analysis, earnings data, and layer-level priors. Companies are ranked by AI Fair Value — the larger your market cap and the higher your AI Revenue %, the higher you rank.
AI Revenue % Methodology
AI Revenue % estimates how much of a company's total revenue comes from AI-related products and services. We update it quarterly using:
- Explicit disclosures — Some companies report AI revenue directly (e.g., Palantir's AIP revenue, SoundHound's AI-native revenue).
- Segment analysis — For companies like Microsoft or Google, we derive AI revenue from segment breakdowns and management commentary (e.g., "AI annual run rate hit $13B").
- SEC filing analysis — We analyze 10-K and 10-Q filings to identify AI-related revenue disclosures, using automated analysis cross-referenced with earnings call transcripts.
Raw estimates are then refined through our Bayesian model, which provides a posterior mean and confidence interval for each company.
How Estimates Work
Rather than treating each AI Revenue % estimate as a fixed number, we model it as a range. Companies with more data get sharper estimates. Companies with less data are pulled toward their layer average until more evidence arrives.
Data Source & Precision
Each company has two quality indicators: source (how the AI revenue data was gathered) and precision (how narrow the confidence range is).
Rank Ranges
Rank ranges show where a company would rank under different scenarios. The range uses the P95 (optimistic) and P5 (conservative) AI Revenue % estimates. A company ranked #25 with range (20-30) could plausibly rank anywhere from #20 to #30 given current uncertainty. Tighter ranges indicate more confident rankings.
Update Cadence
Prices and rankings update daily after market close. AI Revenue % estimates are refreshed quarterly after earnings season. Layer classifications are updated as needed.
The 9-Layer Framework
We classify companies by their primary role in the AI value chain. Some companies span multiple layers (e.g., Google does cloud compute AND builds foundation models). We assign the layer that represents their primary AI revenue driver.
Raw Materials & Mining
Companies extracting and processing rare earth elements, silicon, and other materials essential for semiconductor manufacturing.
Semiconductor Equipment
Manufacturers of the machines used to fabricate chips — lithography, etching, deposition, and inspection systems.
Chip Design & Fabrication
Companies designing and manufacturing GPUs, AI accelerators, and custom silicon that power AI workloads.
Hardware & Systems
Server manufacturers, networking equipment, and power/cooling infrastructure for AI data centers.
Cloud & Compute
Hyperscale cloud providers and GPU cloud platforms delivering on-demand AI compute infrastructure.
Data Infrastructure
Databases, data warehouses, and data pipeline companies enabling the storage and processing of AI training data.
AI Frameworks & Models
Companies building foundation models, training frameworks, and core AI research. Many are still private.
AI Software & Platforms
Enterprise AI platforms, MLOps tools, and AI-powered software infrastructure companies.
AI Applications & Verticals
Companies deploying AI in specific industries — autonomous vehicles, healthcare, robotics, and more.
Data Sources
Market data is sourced from Finnhub. Financial statements and balance sheet data come from SEC EDGAR XBRL filings. AI revenue estimates are derived from SEC filings (10-K, 10-Q), earnings call transcripts, and LLM-powered analysis (Claude by Anthropic). Company classifications are maintained editorially. This is not financial advice.
Disclaimers
Not investment advice. canar.ai is for informational purposes only. Nothing on this site constitutes investment advice, a recommendation, or a solicitation to buy or sell any security. Always do your own research and consult a qualified financial advisor before making investment decisions.
AI Revenue % is model-generated. Most companies do not report AI revenue as a standalone line item. Our estimates are produced by large language models analyzing SEC filings and earnings data, refined through Bayesian estimation, then validated by editorial review. The Bayesian model provides confidence intervals that reflect this uncertainty.
Market-cap dependency. AI Fair Value is Market Cap × AI Revenue %, so rankings will move with stock prices even if the underlying AI business hasn't changed. This is by design — we want to reflect the market's current valuation of each company's AI business.
LLM estimation disclosure. AI Revenue % estimates are generated by Claude (Anthropic) analyzing SEC filings and publicly available earnings data. These estimates may contain errors or misinterpretations of financial disclosures. All estimates are subject to editorial review but should not be treated as authoritative financial data.