# Next Batch Plan

Status: PASS

## Selected Companies

The next three companies after DLR should be **EQIX**, **APLD**, and **CRWV**.

- **EQIX** is the nearest public comparable to DLR. It should test whether the DLR extraction pattern generalizes to a second data-center REIT with quarterly result packages.
- **APLD** is the best near-term AI/HPC infrastructure developer for power/campus forecast-vs-actual tracking. Its SEC filings and press releases are likely to expose secured power, project status, and customer commitments.
- **CRWV** is the best AI cloud demand bridge. It has shorter public history, but RPO/backlog, capex, and customer commitments are directly relevant to AI infrastructure demand conversion.

## Why Not Start With Hyperscalers

ORCL, MSFT, AMZN, GOOGL, and META are critical, but they usually disclose corporate capex, cloud revenue, and qualitative AI infrastructure commentary rather than physical capacity. They should enter after the row schema and validation protocol is proven on companies with more direct facility/power disclosure.

## First Task To Run Next

The single task should run next is: **build an EQIX quarterly-results source manifest and extractor for capex actuals, development/pipeline capacity, and revenue conversion across the latest eight quarters.**

Reason: EQIX is closest to DLR in business model and disclosure style, so it is the highest-signal test of whether the DLR extractor approach can generalize without changing the canonical schema.

## Acceptance For That Next Task

- Produce `eqix_source_manifest.csv`, `extract_eqix_quarterly.py`, `eqix_rows.csv`, `validate_eqix_quarterly.py`, and `eqix_verification_report.md`.
- Validator must compile and exit 0.
- Rows must include primary-source snippets and explicit caveats where EQIX reports capacity in non-MW units.
- Reviewer must fail any silent conversion across USD capex, physical capacity, and revenue.