Methodology

Methodology

Our coverage rules and data pipeline, made explicit so you can audit every claim.

5-Layer AI Value Chain

Every AI-tagged ticker is mapped to one or more layers below. Multi-layer tickers (e.g. NVDA = Chip + Infra) reflect business-line breadth.
Chip
Semiconductors, GPUs, accelerators, fab equipment. Examples: NVDA, AMD, AVGO, ASML, TSM.
Infra
Cloud compute, networking, hyperscaler capex. Examples: MSFT, AMZN, GOOGL (Cloud segment), ANET, CRWV.
Model
Foundation-model labs and research-led product. Examples: GOOGL (DeepMind), META (Llama), MSFT (OpenAI partnership).
App
Productized AI surfaces — consumer or enterprise. Examples: AAPL (Apple Intelligence), CRM, NOW, TEAM.
Enabler
Power, real estate, cooling, materials enabling AI infra. Examples: VST, CEG, EQIX, CCJ.

Data Sources

  • SEC EDGAR 13F-HR — Berkshire Hathaway 13F-HR via SEC EDGAR (CIK 0001067983) using edgartools 5.35. Refreshed via cron daily at 06:00 UTC.
  • ARK Invest EOD CSV — ARK Investment Management 13F-HR via SEC EDGAR (CIK 0001697748) + daily holdings CSV from ARK Invest's official disclosure URL. Refreshed via cron daily at 03:00 UTC Tue-Sat.

Quote Rules (every quote satisfies all four)

  1. 1. Source URL that is publicly clickable and stable.
  2. 2. Human-verified original text by a named curator.
  3. 3. Archived excerpt of the original passage stored against link rot.
  4. 4. Curator signs their real name to the entry.