#!/usr/bin/env python3 """ Manifest-driven APLD forecast-vs-actual ledger extractor. The extractor intentionally uses primary-source manifest entries plus compact source fragments and regex extraction rules. It does not scrape prose summaries. Run from the artifacts directory: python extract_apld_forecast_actual.py Outputs: apld_forecast_actual_rows.csv """ from __future__ import annotations import csv import re from decimal import Decimal from pathlib import Path from typing import Dict, Iterable, List OUT = Path(__file__).resolve().parent MANIFEST_PATH = OUT / "apld_forecast_actual_manifest.csv" ROWS_PATH = OUT / "apld_forecast_actual_rows.csv" FIELDS = [ "company", "period", "metric_name", "raw_value", "normalized_value", "unit", "value_qualifier", "ownership_basis", "forecast_or_actual", "is_derived", "source_url", "source_location", "source_snippet", "snippet_support_status", "confidence", "caveat", ] ROW_SPECS = [ { "source_id": "APLD_2025_10K", "company": "APLD", "period": "FY2025", "metric_name": "capex_actual_total", "raw_value": "$985.826 million", "unit": "USD_millions", "value_qualifier": "exact", "ownership_basis": "consolidated", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-K Note 16 segment disclosures, lines 3889-3901", "source_snippet": "Total capital expenditures $985,826", "extract_regex": r"Total capital expenditures\s*\$?([0-9,]+)", "scale": "0.001", "confidence": "high", "caveat": "SEC table is in thousands; normalized to USD millions.", }, { "source_id": "APLD_2025_10K", "company": "APLD", "period": "FY2025", "metric_name": "capex_actual_hpc_hosting", "raw_value": "$976.506 million", "unit": "USD_millions", "value_qualifier": "exact", "ownership_basis": "HPC Hosting segment", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-K Note 16 segment disclosures, lines 3889-3898", "source_snippet": "HPC Hosting Business 976,506", "extract_regex": r"HPC Hosting Business\s*\$?([0-9,]+)", "scale": "0.001", "confidence": "high", "caveat": "SEC table is in thousands; segment capex includes HPC Hosting Business.", }, { "source_id": "APLD_2025_10K", "company": "APLD", "period": "FY2025", "metric_name": "construction_in_progress", "raw_value": "$1.123156 billion", "unit": "USD_millions", "value_qualifier": "exact", "ownership_basis": "consolidated property and equipment", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-K Note 3 property and equipment, lines 2673-2703", "source_snippet": "Construction in progress 1,123,156", "extract_regex": r"Construction in progress\s*\$?([0-9,]+)", "scale": "0.001", "confidence": "high", "caveat": "SEC table is in thousands; includes assets received but not placed in service.", }, { "source_id": "APLD_2026Q1_10Q", "company": "APLD", "period": "FY2026Q1", "metric_name": "capex_actual_purchases_property_equipment_other_assets", "raw_value": "$249.420 million", "unit": "USD_millions", "value_qualifier": "exact", "ownership_basis": "consolidated cash-flow statement", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-Q cash flows, lines 429-430", "source_snippet": "Purchases ... property and equipment ... (249,420)", "extract_regex": r"\(([0-9,]+)\)", "scale": "0.001", "confidence": "high", "caveat": "Cash-flow line includes other assets and is presented in thousands.", }, { "source_id": "APLD_2026Q1_10Q", "company": "APLD", "period": "FY2026Q1", "metric_name": "construction_in_progress", "raw_value": "$1.309756 billion", "unit": "USD_millions", "value_qualifier": "exact", "ownership_basis": "consolidated property and equipment", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-Q Note 3 property and equipment, lines 607-636", "source_snippet": "Construction in progress 1,309,756", "extract_regex": r"Construction in progress\s*\$?([0-9,]+)", "scale": "0.001", "confidence": "high", "caveat": "SEC table is in thousands.", }, { "source_id": "APLD_2026Q1_10Q", "company": "APLD", "period": "FY2026Q1", "metric_name": "live_capacity_data_center_hosting", "raw_value": "106 MW Jamestown + 180 MW Ellendale", "unit": "MW", "value_qualifier": "exact_sum", "ownership_basis": "operated data-center hosting facilities", "forecast_or_actual": "actual", "is_derived": "true", "source_location": "10-Q business update, lines 1593-1598", "source_snippet": "106 MW ... 180 MW ... full capacity", "extract_regex": r"([0-9]+)\s*MW.*?([0-9]+)\s*MW", "derive": "sum_groups", "confidence": "high", "caveat": "Derived by summing two operated crypto data-center hosting facilities.", }, { "source_id": "APLD_2026Q1_10Q", "company": "APLD", "period": "FY2026Q1", "metric_name": "contracted_capacity_polaris_forge_1", "raw_value": "400 MW", "unit": "MW", "value_qualifier": "exact", "ownership_basis": "APLD leased critical IT load", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-Q leases/business update, lines 1148-1164 and 1601-1606", "source_snippet": "total capacity under contract ... 400 MW", "extract_regex": r"([0-9]+)\s*MW", "confidence": "high", "caveat": "Represents contracted critical IT capacity, not fully energized capacity.", }, { "source_id": "APLD_2026Q1_10Q", "company": "APLD", "period": "FY2026Q1", "metric_name": "minimum_contracted_lease_payments", "raw_value": "$10.851517 billion", "unit": "USD_millions", "value_qualifier": "exact", "ownership_basis": "lessor minimum contracted payments", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-Q Note 11 leases, lines 1164-1172", "source_snippet": "Total $10,851,517", "extract_regex": r"Total\s*\$?([0-9,]+)", "scale": "0.001", "confidence": "high", "caveat": "SEC table is in thousands and excludes operating-expense reimbursements and variable rent increases.", }, { "source_id": "APLD_2025_06_02_PR", "company": "APLD", "period": "CY2025-06-02", "metric_name": "contracted_capacity_coreweave_initial", "raw_value": "250 MW", "unit": "MW", "value_qualifier": "up_to", "ownership_basis": "APLD leased critical IT load", "forecast_or_actual": "forecast", "is_derived": "false", "source_location": "IR press release PDF, page 1 lines 4-15", "source_snippet": "deliver 250 megawatts", "extract_regex": r"([0-9]+)\s*megawatts", "confidence": "high", "caveat": "Forward-looking lease delivery statement at announcement date.", }, { "source_id": "APLD_2025_06_02_PR", "company": "APLD", "period": "CY2025-06-02", "metric_name": "contracted_revenue_coreweave_initial", "raw_value": "approximately $7 billion", "unit": "USD_millions", "value_qualifier": "approximately", "ownership_basis": "APLD anticipated lease revenue", "forecast_or_actual": "forecast", "is_derived": "false", "source_location": "IR press release PDF, page 1 lines 8-12", "source_snippet": "approximately $7 billion", "extract_regex": r"\$([0-9]+)\s*billion", "scale": "1000", "confidence": "high", "caveat": "Management anticipated total revenue over approximately 15-year lease terms.", }, { "source_id": "APLD_2025_06_02_PR", "company": "APLD", "period": "CY2025-06-02", "metric_name": "expected_commencement_window_building_2", "raw_value": "ready for service in fourth quarter calendar 2025", "normalized_value": "2025-Q4", "unit": "calendar_quarter", "value_qualifier": "expected", "ownership_basis": "APLD leased facility schedule", "forecast_or_actual": "forecast", "is_derived": "false", "source_location": "IR press release PDF, page 2 lines 35-36", "source_snippet": "100 MW ... fourth quarter ... 2025", "extract_regex": r"100\s*MW.*fourth quarter.*2025", "confidence": "high", "caveat": "Forward-looking ready-for-service timing.", }, { "source_id": "APLD_2025_06_02_PR", "company": "APLD", "period": "CY2025-06-02", "metric_name": "expected_commencement_window_building_3", "raw_value": "150 MW data center ready for service in middle of 2026", "normalized_value": "2026-mid", "unit": "calendar_window", "value_qualifier": "expected", "ownership_basis": "APLD leased facility schedule", "forecast_or_actual": "forecast", "is_derived": "false", "source_location": "IR press release PDF, page 2 lines 36-38", "source_snippet": "150 MW ... middle of 2026", "extract_regex": r"150\s*MW.*middle of 2026", "confidence": "high", "caveat": "Forward-looking schedule; not a precise quarter.", }, { "source_id": "APLD_2025_08_29_8K", "company": "APLD", "period": "CY2025-08-29", "metric_name": "contracted_capacity_building_4", "raw_value": "150 MW", "unit": "MW", "value_qualifier": "exact", "ownership_basis": "APLD leased critical IT load", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "8-K Item 1.01, lines 54-58", "source_snippet": "Building 4 ... 150MW", "extract_regex": r"([0-9]+)\s*MW", "confidence": "high", "caveat": "Actual executed lease; capacity delivery remains future construction.", }, { "source_id": "APLD_2025_08_29_8K", "company": "APLD", "period": "CY2025-08-29", "metric_name": "contracted_revenue_building_4", "raw_value": "approximately $4.0 billion", "unit": "USD_millions", "value_qualifier": "approximately", "ownership_basis": "APLD anticipated lease revenue", "forecast_or_actual": "forecast", "is_derived": "false", "source_location": "8-K Item 1.01, lines 56-58", "source_snippet": "approximately $4.0 billion", "extract_regex": r"\$([0-9.]+)\s*billion", "scale": "1000", "confidence": "high", "caveat": "Expected contract value for initial approximately 15-year term.", }, { "source_id": "APLD_2026Q2_10Q", "company": "APLD", "period": "FY2026Q2", "metric_name": "capacity_added_this_quarter_hpc", "raw_value": "100 MW completed and operational", "unit": "MW", "value_qualifier": "exact", "ownership_basis": "APLD Polaris Forge 1 ELN-02", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-Q lease note/business update, lines 1813-1816 and 2311-2314", "source_snippet": "100 MW ... completed and operational", "extract_regex": r"([0-9]+)\s*MW", "confidence": "medium", "caveat": "Filing has October/November wording in separate sections; quarter-level ledger treats it as FY2026Q2 capacity addition.", }, { "source_id": "APLD_2026Q2_10Q", "company": "APLD", "period": "FY2026Q2", "metric_name": "revenue_conversion_hpc_total", "raw_value": "$85.0 million HPC Hosting revenue", "unit": "USD_millions", "value_qualifier": "approximately", "ownership_basis": "HPC Hosting segment", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-Q business update and MD&A, lines 2321-2323 and 2566-2567", "source_snippet": "$85.0 million ... HPC Hosting", "extract_regex": r"\$([0-9.]+)\s*million", "confidence": "high", "caveat": "Segment revenue includes tenant fit-out and rental revenue.", }, { "source_id": "APLD_2026Q2_10Q", "company": "APLD", "period": "FY2026Q2", "metric_name": "revenue_conversion_tenant_fitout", "raw_value": "$73.0 million tenant fit-out services", "unit": "USD_millions", "value_qualifier": "approximately", "ownership_basis": "HPC Hosting segment", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-Q MD&A, lines 2566-2570", "source_snippet": "$73.0 million ... tenant fit-out", "extract_regex": r"\$([0-9.]+)\s*million", "confidence": "high", "caveat": "Tenant fit-out revenue is not recurring rental revenue.", }, { "source_id": "APLD_2026Q2_10Q", "company": "APLD", "period": "FY2026Q2", "metric_name": "revenue_conversion_rental", "raw_value": "$12.0 million rental revenues", "unit": "USD_millions", "value_qualifier": "approximately", "ownership_basis": "Polaris Forge 1 ELN-02", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-Q MD&A, lines 2566-2567", "source_snippet": "$12.0 million ... rental revenues", "extract_regex": r"\$([0-9.]+)\s*million", "confidence": "high", "caveat": "Rental revenue appeared as ELN-02 fully energized during the quarter.", }, { "source_id": "APLD_2026Q2_10Q", "company": "APLD", "period": "FY2026Q2", "metric_name": "minimum_contracted_lease_payments", "raw_value": "$15.551105 billion", "unit": "USD_millions", "value_qualifier": "exact", "ownership_basis": "lessor minimum contracted payments", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-Q Note 11 leases, lines 1817-1839", "source_snippet": "Total $15,551,105", "extract_regex": r"Total\s*\$?([0-9,]+)", "scale": "0.001", "confidence": "high", "caveat": "SEC table is in thousands and excludes operating-expense reimbursements and variable rent increases.", }, { "source_id": "APLD_2026Q2_10Q", "company": "APLD", "period": "FY2026Q2", "metric_name": "contracted_capacity_polaris_forge_2", "raw_value": "200 MW", "unit": "MW", "value_qualifier": "exact", "ownership_basis": "APLD leased critical IT load", "forecast_or_actual": "actual", "is_derived": "false", "source_location": "10-Q lease note/business update, lines 1816 and 2321-2322", "source_snippet": "lease ... 200MW ... Polaris Forge 2", "extract_regex": r"([0-9]+)\s*MW", "confidence": "high", "caveat": "Executed lease commitment; facility remains under construction.", }, { "source_id": "APLD_2026Q2_10Q", "company": "APLD", "period": "FY2026Q2", "metric_name": "expected_commencement_window_polaris_forge_2", "raw_value": "initial capacity in 2026 and full capacity in early 2027", "normalized_value": "initial:2026;full:2027-early", "unit": "calendar_window", "value_qualifier": "expected", "ownership_basis": "APLD Polaris Forge 2 project schedule", "forecast_or_actual": "forecast", "is_derived": "false", "source_location": "10-Q business update, lines 2321-2322", "source_snippet": "initial capacity in 2026 ... full ... early 2027", "extract_regex": r"initial capacity in 2026.*early 2027", "confidence": "high", "caveat": "Forward-looking project schedule.", }, ] def load_manifest(path: Path) -> Dict[str, dict]: with path.open(newline="", encoding="utf-8") as f: return {row["source_id"]: row for row in csv.DictReader(f)} def decimal_from_match(match: re.Match[str], spec: dict) -> str: if spec.get("derive") == "sum_groups": value = sum(Decimal(g.replace(",", "")) for g in match.groups()) else: value = Decimal(match.group(1).replace(",", "")) if "scale" in spec: value *= Decimal(spec["scale"]) return format(value.normalize(), "f") def extract_rows(specs: Iterable[dict], manifest: Dict[str, dict]) -> List[dict]: rows: List[dict] = [] for spec in specs: source = manifest[spec["source_id"]] snippet = spec["source_snippet"] pattern = spec["extract_regex"] match = re.search(pattern, snippet, flags=re.IGNORECASE | re.DOTALL) support = "supported_regex_match" if match else "unsupported_no_regex_match" normalized_value = spec.get("normalized_value") if normalized_value is None: normalized_value = decimal_from_match(match, spec) if match else "" row = {field: "" for field in FIELDS} for field in FIELDS: if field in spec: row[field] = str(spec[field]) row["normalized_value"] = str(normalized_value) row["source_url"] = source["source_url"] row["snippet_support_status"] = support rows.append(row) return rows def write_rows(rows: List[dict], path: Path) -> None: with path.open("w", newline="", encoding="utf-8") as f: writer = csv.DictWriter(f, fieldnames=FIELDS) writer.writeheader() writer.writerows(rows) def main() -> None: manifest = load_manifest(MANIFEST_PATH) rows = extract_rows(ROW_SPECS, manifest) write_rows(rows, ROWS_PATH) print(f"wrote {len(rows)} rows to {ROWS_PATH}") if __name__ == "__main__": main()