Data

Download every dataset, with its fields, terms and dates

Every table on this site as CSV and as JSON, with the day its data was last checked and the day it is due again.

catalogue.json the whole catalogue, without the rows API each dataset at a fixed address, with field types that stay fixed Releases a fixed copy of each file every time it changed Status each dataset’s state and its source’s next release Reports each quarter in six findings, fixed, with a PDF Answers one number for each common question, from these datasets

Buildout, Power and others are due for a refresh

9 Oct 2026 to 11 Dec 2026
  1. Buildout 3 files, 892 rows 11 Nov 2026
  2. Power 10 files, 2,203 rows 10 Nov 2026
  3. Machine 5 files, 4,190 rows 13 Nov 2026
  4. Chips 14 files, 4,930 rows 13 Nov 2026
  5. Jobs 1 file, 832 rows 10 Nov 2026
  6. Ledger 1 file, 63 rows 20 Nov 2026
  7. Frontier 6 files, 978 rows 6 Nov 2026
  8. Vendors 5 files, 1,012 rows 8 Nov 2026
  9. Rules 2 files, 81 rows 26 Nov 2026
  10. Asia 1 file, 25 rows 11 Dec 2026
  11. Dispatch 1 file, 5 rows on demand
  12. Records 22 files, 3,108 rows 10 Nov 2026
  13. Grapher 1 file, 3,906 rows 10 Nov 2026
  14. Companies 1 file, 18 rows 6 Nov 2026
  15. Explorer 3 files, 602 rows 10 Nov 2026
Each bar runs from the day the section’s data was last checked to the first day any of it is due again, the date at its right. The upright line is the day this page was built, 12 Nov 2026.

Buildout

Buildout →
Dataset Rows Terms Checked Due by Files
AI campuses, planned against reached One row per planned step.
18 fields and code
  • campus text
  • owner text
  • planned_in_release text
  • planned_date text
  • power_before_mw number
  • planned_power_mw number
  • step_mw number
  • due text
  • newest_release text
  • newest_power_on_planned_date_mw number
  • reached text
  • reached_is text
  • days_late number
  • short_of_plan_mw number
  • outcome text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/buildout/promises.csv")
df.shape  # (91, 18)
 91 CC BY 4.0 22 Oct 2026 26 Nov 2026 CSV JSON Releases
AI data centers One row per campus.
19 fields and code
  • slug text
  • name text
  • owner text
  • users text
  • country text
  • place text
  • town_lat number
  • town_lon number
  • facility_mw number
  • it_mw number
  • h100e number
  • capex_usd_billions number
  • as_of text
  • first_powered text
  • peak_facility_mw number
  • peak_date text
  • chips text
  • page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/buildout/sites.csv")
df.shape  # (95, 19)
 95 CC BY 4.0 22 Oct 2026 26 Nov 2026 CSV JSON Releases
US data-center construction spending, every release One row per month in each kept release.
10 fields and code
  • release text
  • month text
  • data_center_usd_millions_saar number
  • status text
  • change_on_year_earlier number
  • newest_release text
  • table text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/buildout/spending.csv")
df.shape  # (706, 10)
 706 Public domain 16 Oct 2026 11 Nov 2026 CSV JSON Releases

Power

Power →
Dataset Rows Terms Checked Due by Files
Commercial and industrial power prices, and a year’s bill One row per state, utility or site.
15 fields and code
  • kind text
  • id text
  • name text
  • state text
  • detail text
  • industrial_cents_per_kwh number
  • commercial_cents_per_kwh number
  • industrial_usd_per_mw_year number
  • commercial_usd_per_mw_year number
  • mw number
  • usd_per_year_at_full_draw number
  • period text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/power/bill.csv")
df.shape  # (233, 15)
 233 public domain; Epoch AI data CC BY 4.0 22 Oct 2026 20 Nov 2026 CSV JSON Releases
PJM capacity price by zone One row per zone and delivery year.
17 fields and code
  • zone text
  • name text
  • part text
  • states text
  • delivery_year text
  • start text
  • end text
  • priced_as text
  • price_usd_per_mw_day number
  • fpr number
  • days number
  • usd_per_kw_month number
  • report text
  • report_page number
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/power/capacity.csv")
df.shape  # (110, 17)
 110 figures cited from its reports 16 Oct 2026 14 Jan 2027 CSV JSON Releases
Fuel mix of US grids with AI sites One row per grid.
19 fields and code
  • grid_code text
  • grid text
  • months_from text
  • months_to text
  • months_reported number
  • net_generation_mwh number
  • share_gas number
  • share_coal number
  • share_nuclear number
  • share_wind number
  • share_solar number
  • share_hydro number
  • share_other number
  • fuel_hours_share number
  • ai_counties number
  • ai_sites number
  • ai_facility_mw_2026-10-22 number
  • ai_site_pages text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/power/grids.csv")
df.shape  # (25, 19)
 25 Public domain; CC BY 4.0 16 Oct 2026 20 Nov 2026 CSV JSON Releases
Nuclear power agreements for AI One row per agreement.
19 fields and code
  • id text
  • announced text
  • buyer text
  • counterparty text
  • project text
  • location text
  • kind text
  • status text
  • mw number
  • mw_qualifier text
  • counted_mw number
  • part_of text
  • first_power number
  • full_by number
  • term_years number
  • releases text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/power/nuclear.csv")
df.shape  # (22, 19)
 22 facts, quoted with credit 11 Oct 2026 10 Nov 2026 CSV JSON Releases
Power plants planned, added and retired near AI sites One row per generator and kind of change.
18 fields and code
  • change text
  • plant_id number
  • plant text
  • owner text
  • state text
  • county text
  • generator_id text
  • group text
  • technology text
  • energy_source text
  • nameplate_mw number
  • status text
  • status_text text
  • date text
  • ai_sites text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/power/plants.csv")
df.shape  # (461, 18)
 461 public domain 30 Oct 2026 26 Nov 2026 CSV JSON Releases
ERCOT large-load queue: MW by ERCOT’s status words and by year, from each monthly operational overview One row per status and year in one monthly overview.
13 fields and code
  • month text
  • status text
  • year number
  • printed number
  • unit text
  • mw number
  • overview text
  • url text
  • page number
  • pdf_sha256 text
  • table_image_sha256 text
  • publisher text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/power/queue.csv")
df.shape  # (957, 13)
 957 ERCOT terms of use: public contents may be reproduced unmodified, and raw data used in compilations, charts and analyses (https://www.ercot.com/help/terms) 3 Nov 2026 18 Dec 2026 CSV JSON Releases
Power prices and AI capacity by state One row per state.
16 fields and code
  • state text
  • code text
  • home_cents_per_kwh_2021-01_2021-12 number
  • home_cents_per_kwh_2025-08_2026-07 number
  • home_change number
  • all_cents_per_kwh_2021-01_2021-12 number
  • all_cents_per_kwh_2025-08_2026-07 number
  • all_change number
  • average_demand_mw number
  • ai_facility_mw_2026-10-22 number
  • ai_share_of_demand number
  • ai_facility_mw_2030-01-01_projected number
  • ai_sites number
  • eia_status text
  • page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/power/states.csv")
df.shape  # (51, 16)
 51 Public domain; CC BY 4.0 22 Oct 2026 20 Nov 2026 CSV JSON Releases
Large-load tariffs and rules for data centers One row per tariff or rule.
23 fields and code
  • id text
  • kind text
  • utility text
  • parent text
  • states text
  • regulator text
  • docket text
  • name text
  • status text
  • filed text
  • decided text
  • threshold_mw number
  • threshold_note text
  • minimum_bill_pct number
  • term_years number
  • ramp_years number
  • exit_fee text
  • collateral text
  • eia_ids text
  • sources text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/power/tariffs.csv")
df.shape  # (44, 23)
 44 facts, quoted with credit 19 Oct 2026 18 Nov 2026 CSV JSON Releases
Utilities serving counties with AI sites One row per utility in a county.
14 fields and code
  • state text
  • county text
  • county_geoid text
  • ai_sites text
  • ai_facility_mw_2026-10-22 number
  • eia_utility_number number
  • utility text
  • ownership text
  • home_cents_per_kwh_2021 number
  • home_cents_per_kwh_2025 number
  • home_change number
  • home_customers_2025 number
  • page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/power/utilities.csv")
df.shape  # (245, 14)
 245 Public domain; CC BY 4.0 18 Oct 2026 15 Feb 2027 CSV JSON Releases
Power mix of every country that hosts AI One row per country.
25 fields and code
  • code text
  • iso3 text
  • country text
  • period_kind text
  • period_from text
  • period_to text
  • twh number
  • share_gas number
  • share_coal number
  • share_oil number
  • share_nuclear number
  • share_wind number
  • share_solar number
  • share_hydro number
  • share_bio number
  • fossil number
  • nuclear number
  • renewables number
  • intensity_g_per_kwh number
  • fossil_change number
  • ai_mw_now number
  • ai_mw_end number
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/power/world.csv")
df.shape  # (55, 25)
 55 CC BY 4.0 11 Oct 2026 15 Nov 2026 CSV JSON Releases

Machine

Machine →
Dataset Rows Terms Checked Due by Files
Chip plants One row per plant.
11 fields and code
  • company text
  • plant text
  • place text
  • country text
  • status text
  • wafer_300mm text
  • latitude number
  • longitude number
  • page text
  • license text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/machine/chip-plants.csv")
df.shape  # (456, 11)
 456 CC BY-SA 4.0 22 Oct 2026 21 Dec 2026 CSV JSON Releases
Chip prices, output and parts One row per reading of a series.
12 fields and code
  • source text
  • measure text
  • series text
  • period text
  • value number
  • low_5th number
  • high_95th number
  • unit text
  • preliminary text
  • page text
  • source_url text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/machine/chips.csv")
df.shape  # (2022, 12)
 2,022 public domain; CC BY 4.0 11 Oct 2026 15 Nov 2026 CSV JSON Releases
US imports of servers and chips One row per month, country and HS code.
9 fields and code
  • month text
  • country text
  • census_country_code number
  • hs6 number
  • goods text
  • general_imports_usd number
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/machine/imports.csv")
df.shape  # (1560, 9)
 1,560 Public domain 10 Oct 2026 14 Nov 2026 CSV JSON Releases
Monthly sales of Taiwan’s AI hardware makers One row per company or group and month.
14 fields and code
  • kind text
  • code number
  • company text
  • company_zh text
  • month text
  • revenue_ntd_thousand number
  • same_month_year_earlier number
  • growth_pct number
  • note_zh text
  • note_en text
  • matched text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/machine/sales.csv")
df.shape  # (78, 14)
 78 Open Government Data License, version 1.0 30 Oct 2026 13 Nov 2026 CSV JSON Releases
AI computing power by site One row per site.
10 fields and code
  • name text
  • owner text
  • country text
  • h100_equivalents number
  • as_of text
  • first_power text
  • latitude number
  • longitude number
  • page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/machine/supercomputers.csv")
df.shape  # (74, 10)
 74 CC BY 4.0 22 Oct 2026 26 Nov 2026 CSV JSON Releases

Chips

Chips →
Dataset Rows Terms Checked Due by Files
The accelerator record: each AI chip’s figures and dates, quoted One row per figure, price, promise or shipment of one chip.
21 fields and code
  • kind text
  • vendor text
  • chip text
  • field text
  • value number
  • unit text
  • printed text
  • printed_unit text
  • basis text
  • per_chips number
  • stated text
  • target text
  • window_end text
  • shipped text
  • precision text
  • status text
  • slip_months_at_most number
  • quote text
  • source_url text
  • license text
  • page text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/accelerators.csv")
df.shape  # (178, 21)
 178 quoted for comment 23 Oct 2026 22 Dec 2026 CSV JSON Releases
Chip makers’ books: capital spending, research, inventory and sales to China One row per company and fiscal year, or company and place.
20 fields and code
  • kind text
  • company text
  • group text
  • fiscal_year_end text
  • currency text
  • revenue number
  • capex number
  • rd number
  • inventory number
  • cost_of_revenue number
  • capex_share number
  • rd_share number
  • inventory_days number
  • place text
  • place_revenue number
  • place_share number
  • accession text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/books.csv")
df.shape  # (423, 20)
 423 public domain 22 Oct 2026 6 Dec 2026 CSV JSON Releases
The US Entity List at every version, and BIS chip rules One row per version, year, country or rule.
15 fields and code
  • kind text
  • date text
  • country text
  • entries number
  • china number
  • russia number
  • added number
  • removed number
  • fdp_entries number
  • title text
  • citation text
  • url text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/controls.csv")
df.shape  # (261, 15)
 261 public domain 19 Oct 2026 18 Nov 2026 CSV JSON Releases
Die and package cost lab One row per die size, yield case, chiplet split, interposer or chip.
17 fields and code
  • kind text
  • id text
  • label text
  • area_mm2 number
  • whole_per_wafer number
  • formula_per_wafer number
  • defects_per_cm2 number
  • yield_model text
  • yield number
  • good_per_wafer number
  • wafer_usd number
  • usd_each number
  • chiplets number
  • extra_area number
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/cost.csv")
df.shape  # (115, 17)
 115 CC BY 4.0 19 Oct 2026 3 Dec 2026 CSV JSON Releases
The chip cycle board: ten monthly series, how many improve, and which lead One row per series month, diffusion month or series lead.
13 fields and code
  • kind text
  • series text
  • month text
  • change_on_a_year number
  • unit text
  • improving number
  • of_series number
  • lead_months number
  • correlation number
  • months_compared number
  • source text
  • license text
  • page text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/cycle.csv")
df.shape  # (629, 13)
 629 public domain; UN Comtrade, reuse with credit; public domain and UN Comtrade, as each series 19 Oct 2026 23 Nov 2026 CSV JSON Releases
Construction spending at US chip, electronics and electrical plants, by month, with data centers One row per month.
11 fields and code
  • month text
  • computer_electronic_electrical_musd_saar number
  • flag text
  • yoy_change number
  • share_of_manufacturing number
  • share_of_nonresidential number
  • data_center_musd_saar number
  • release text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/fab-spending.csv")
df.shape  # (404, 11)
 404 Public domain 20 Oct 2026 15 Nov 2026 CSV JSON Releases
The fab record: each plant’s dated fields and public money, quoted One row per one dated field of one plant.
15 fields and code
  • field text
  • plant_id text
  • owner text
  • plant text
  • country text
  • date text
  • target text
  • window_end text
  • value text
  • unit text
  • detail text
  • quote text
  • source_url text
  • license text
  • page text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/fabs.csv")
df.shape  # (130, 15)
 130 quoted for comment 20 Oct 2026 4 Dec 2026 CSV JSON Releases
AI chips on the inference benchmark: each chip’s best result per chip in two compatible rounds One row per chip, benchmark and scenario in two rounds.
25 fields and code
  • benchmark text
  • benchmark_name text
  • scenario text
  • units text
  • chip text
  • from_round number
  • from_result_id text
  • from_submitter text
  • from_system text
  • from_chips number
  • from_result number
  • from_per_chip number
  • to_round number
  • to_result_id text
  • to_submitter text
  • to_system text
  • to_chips number
  • to_result number
  • to_per_chip number
  • ratio number
  • compatibility_row text
  • per_chip_status text
  • page text
  • license text
  • trademark text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/inference-gains.csv")
df.shape  # (64, 25)
 64 MLCommons results repository, Apache License 2.0 10 Nov 2026 25 Dec 2026 CSV JSON Releases
AI chips on the inference benchmark: each MLPerf® Inference: Datacenter result, with this site’s figure per chip One row per result of one system in one scenario of one benchmark.
23 fields and code
  • round number
  • result_id text
  • submitter text
  • system text
  • accelerator text
  • chip text
  • chips number
  • chips_note text
  • benchmark text
  • benchmark_name text
  • scenario text
  • result number
  • units text
  • status text
  • per_chip number
  • per_chip_status text
  • published text
  • source_url text
  • kept_copy text
  • kept_sha256 text
  • page text
  • license text
  • trademark text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/inference.csv")
df.shape  # (1441, 23)
 1,441 MLCommons results repository, Apache License 2.0 10 Nov 2026 25 Dec 2026 CSV JSON Releases
The market desk: chip readings furthest from normal, and when this site first held each release’s reading One row per chip cycle series, or one reading of a scheduled release.
21 fields and code
  • kind text
  • rank number
  • series text
  • period text
  • value number
  • change_on_a_year number
  • change_unit text
  • z number
  • release text
  • release_date text
  • release_at_utc text
  • seen_utc text
  • how text
  • commit text
  • lag_minutes number
  • lag_days number
  • press_title text
  • press_url text
  • press_published text
  • page text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/market.csv")
df.shape  # (24, 21)
 24 public domain; UN Comtrade, reuse with credit; Figures cited from public reports 19 Oct 2026 23 Nov 2026 CSV JSON Releases
Chip materials: production, prices and export controls One row per material and country, price year, or control.
13 fields and code
  • kind text
  • material text
  • country text
  • year text
  • series text
  • unit text
  • value number
  • usgs_value text
  • share number
  • detail text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/materials.csv")
df.shape  # (290, 13)
 290 public domain; government notice 19 Oct 2026 18 Nov 2026 CSV JSON Releases
Memory chip prices at the EU border, and Micron’s price and bit index from its own words One row per month, country, statement or quarter.
16 fields and code
  • kind text
  • period text
  • product text
  • country text
  • value number
  • pieces number
  • low number
  • high number
  • change_pct number
  • share_pct number
  • words text
  • derived text
  • url text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/memory-prices.csv")
df.shape  # (644, 16)
 644 Eurostat reuse policy; public domain 20 Oct 2026 24 Nov 2026 CSV JSON Releases
The memory desk: makers’ quarters, Taiwan memory sales, Korea’s memory exports One row per maker quarter, company month, export month or exporter year.
16 fields and code
  • kind text
  • who text
  • period_start text
  • period text
  • currency text
  • revenue_or_value number
  • cost_of_revenue number
  • gross_margin number
  • inventory number
  • year_earlier number
  • kilograms number
  • per_kilogram number
  • note text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/memory.csv")
df.shape  # (255, 16)
 255 public domain; OGDL 1.0; reuse with credit 30 Oct 2026 13 Nov 2026 CSV JSON Releases
The process-node record: each promise and delivery, quoted One row per promise or delivery of one node.
18 fields and code
  • kind text
  • company text
  • node text
  • node_kind text
  • generation text
  • stated text
  • target text
  • window_end text
  • delivered text
  • precision text
  • delivered_by text
  • what text
  • status text
  • slip_months_at_most number
  • quote text
  • source_url text
  • license text
  • page text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/chips/nodes.csv")
df.shape  # (72, 18)
 72 quoted for comment 20 Oct 2026 19 Dec 2026 CSV JSON Releases

Jobs

Jobs →
Dataset Rows Terms Checked Due by Files
US occupations and AI exposure One row per occupation.
26 fields and code
  • soc text
  • title text
  • bls_relative_ai_exposure text
  • bls_employment_2025_thousands number
  • bls_employment_2035_projected_thousands number
  • bls_change_2025_2035_projected_pct number
  • bls_median_wage_2025 text
  • bls_typical_education text
  • bls_names_ai text
  • onet_occupations text
  • onet_tasks number
  • tasks_comparable_with_earlier_samples number
  • task_usage_share_2025-01 number
  • task_usage_share_2025-03 number
  • task_usage_share_2025-08 number
  • task_usage_share_2025-11 number
  • task_usage_share_2026-02 number
  • task_usage_share_2026-04 number
  • task_usage_share_2026-05 number
  • task_usage_share_median number
  • task_usage_share_low number
  • task_usage_share_high number
  • indeed_sector text
  • indeed_postings_index_2026-10-02 number
  • page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/jobs/occupations.csv")
df.shape  # (832, 26)
 832 public domain; CC BY 4.0 11 Oct 2026 10 Nov 2026 CSV JSON Releases

Ledger

Ledger →
Dataset Rows Terms Checked Due by Files
Job cuts blamed on AI One row per record.
13 fields and code
  • record text
  • period text
  • ai_job_cuts number
  • all_job_cuts number
  • notices number
  • workers number
  • industry text
  • size_by_public_float text
  • jobs text
  • what_ai_did text
  • note text
  • page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/ledger/ledger.csv")
df.shape  # (63, 13)
 63 Public domain; Copyright Challenger, Gray & Christmas; No licence is stated 16 Oct 2026 20 Nov 2026 CSV JSON Releases

Frontier

Frontier →
Dataset Rows Terms Checked Due by Files
Days until open weights matched each leader One row per closed leader.
11 fields and code
  • arena text
  • closed_leader text
  • organization text
  • took_first_place text
  • status text
  • matched_on text
  • matched_by text
  • left_board_on text
  • days number
  • page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/frontier/catch-up.csv")
df.shape  # (63, 11)
 63 CC BY 4.0 30 Oct 2026 6 Nov 2026 CSV JSON Releases
Benchmark saturation clock One row per benchmark.
12 fields and code
  • benchmark text
  • released text
  • superseded_by text
  • reached_90_on text
  • days_to_90 number
  • first_model_at_90 text
  • its_score number
  • best_score number
  • best_model text
  • days_running number
  • page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/frontier/clock.csv")
df.shape  # (15, 12)
 15 CC BY 4.0 30 Oct 2026 6 Nov 2026 CSV JSON Releases
Open-weight downloads by Frontier lab, read each week from the Hugging Face Hub One row per lab at one weekly read.
13 fields and code
  • date text
  • lab text
  • accounts text
  • models number
  • downloads_30d number
  • downloads_all_time number
  • likes number
  • all_time_rise text
  • rise_since text
  • all_time_falls number
  • page text
  • license text
  • notice text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/frontier/downloads-labs.csv")
df.shape  # (14, 13)
 14 Counts from the Hugging Face Hub API, with a link to each model 3 Nov 2026 13 Nov 2026 CSV JSON Releases
Open-weight downloads: the top models on the Hugging Face Hub and each Frontier lab’s top ten, read each week One row per model at one weekly read.
15 fields and code
  • date text
  • model text
  • account text
  • lab text
  • task text
  • rank number
  • downloads_30d number
  • downloads_all_time number
  • likes number
  • kept_copy text
  • kept_sha256 text
  • source_url text
  • page text
  • license text
  • notice text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/frontier/downloads.csv")
df.shape  # (194, 15)
 194 Counts from the Hugging Face Hub API, with a link to each model 3 Nov 2026 13 Nov 2026 CSV JSON Releases
First place on the Arena leaderboards One row per reign.
10 fields and code
  • arena text
  • model text
  • organization text
  • weights_on_arena text
  • from text
  • to text
  • days number
  • current text
  • model_page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/frontier/reigns.csv")
df.shape  # (81, 10)
 81 CC BY 4.0 30 Oct 2026 6 Nov 2026 CSV JSON Releases
Model retirements: each AI vendor’s deprecation and retirement dates, quoted One row per model on one vendor platform.
27 fields and code
  • id text
  • source text
  • vendor text
  • platform text
  • model text
  • version text
  • provider text
  • state text
  • retire_kind text
  • retire_date text
  • retire_text text
  • deprecated_date text
  • announced_date text
  • released_date text
  • notice_days number
  • notice_from text
  • replacement text
  • earlier_dates text
  • removed_from_page text
  • quote text
  • kept_copy text
  • kept_sha256 text
  • kept_date text
  • source_url text
  • page text
  • license text
  • notice text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/frontier/retirements.csv")
df.shape  # (611, 27)
 611 Facts from the vendor’s deprecation page, with a link to it 26 Oct 2026 9 Nov 2026 CSV JSON Releases

Vendors

Vendors →
Dataset Rows Terms Checked Due by Files
The AI vendor record: what each vendor’s own pages say on training and retention, quoted One row per one field of one vendor.
14 fields and code
  • vendor text
  • product text
  • field text
  • label text
  • value text
  • quote text
  • context text
  • document text
  • source_url text
  • kept_on text
  • snapshot_sha256 text
  • license text
  • page text
  • notice text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/vendors.csv")
df.shape  # (174, 14)
 174 quoted for comment 4 Nov 2026 4 Dec 2026 CSV JSON Releases
Federal AI contracts: obligations by awarding agency and by recipient, for each reading and window One row per one agency or one recipient in one window.
12 fields and code
  • kind text
  • name text
  • code text
  • reading text
  • window text
  • window_start text
  • window_end text
  • obligations_usd number
  • share_of_window number
  • retrieved text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/vendors/federal-by.csv")
df.shape  # (202, 12)
 202 Public domain 31 Oct 2026 5 Dec 2026 CSV JSON Releases
Federal AI contracts: US contract obligations whose record names artificial intelligence, by month One row per one month, both keyword readings.
8 fields and code
  • month text
  • fiscal_year number
  • says_artificial_intelligence_usd number
  • ai_ml_or_language_model_usd number
  • complete text
  • retrieved text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/vendors/federal.csv")
df.shape  # (109, 8)
 109 Public domain 31 Oct 2026 5 Dec 2026 CSV JSON Releases
The AI vendor questionnaire, answered: each vendor’s answer to nine buyer questions, quoted One row per one question to one vendor.
18 fields and code
  • vendor_id text
  • vendor text
  • question_id text
  • question text
  • state text
  • answer text
  • quote text
  • context text
  • source_title text
  • source_url text
  • kept_on text
  • kept_sha256 text
  • note text
  • nist_ai_rmf text
  • nist_ai_600_1 text
  • page text
  • license text
  • notice text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/vendors/questionnaire.csv")
df.shape  # (189, 18)
 189 Answers quote each vendor’s own page; questions cite NIST AI 600-1 and AI RMF 1.0 (public domain in the US) 4 Nov 2026 8 Nov 2026 CSV JSON Releases
AI service reliability: every incident on AI vendors’ status pages, kept One row per incident on one vendor’s status page.
19 fields and code
  • vendor text
  • id text
  • title text
  • status text
  • first_post text
  • resolved_post text
  • last_post text
  • minutes number
  • components text
  • classes text
  • removed_from_feed text
  • vendor_post text
  • kept_copy text
  • kept_sha256 text
  • first_kept text
  • page text
  • source text
  • license text
  • notice text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/vendors/status-incidents.csv")
df.shape  # (338, 19)
 338 Facts from the vendor’s status feed, with a link to its post 1 Nov 2026 8 Nov 2026 CSV JSON Releases

Rules

Rules →
Dataset Rows Terms Checked Due by Files
The AI enforcement record: each FTC, SEC, Justice Department and FCC action that names AI, quoted One row per action of one agency against one party.
22 fields and code
  • id text
  • agency text
  • party text
  • matter text
  • status text
  • first_date text
  • last_date text
  • steps text
  • party_quote text
  • ai_quote text
  • claim_quote text
  • docket text
  • docket_quote text
  • last_step_quote text
  • last_release_title text
  • last_release_url text
  • last_kept_copy text
  • last_kept_sha256 text
  • related text
  • page text
  • license text
  • notice text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/rules/enforcement/enforcement.csv")
df.shape  # (39, 22)
 39 US federal government work; quoted from the agency’s release, with a link to it 3 Nov 2026 3 Dec 2026 CSV JSON Releases
The AI rules calendar: each dated duty of the EU AI Act and US state AI laws, quoted One row per dated duty of one law.
23 fields and code
  • id text
  • jurisdiction text
  • law text
  • law_name text
  • citation text
  • date text
  • status text
  • kind text
  • who text
  • what text
  • date_text text
  • date_basis text
  • superseded_by text
  • moved_to_date text
  • quote text
  • source_title text
  • source_url text
  • kept_copy text
  • kept_sha256 text
  • kept_date text
  • page text
  • license text
  • notice text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/rules/rules.csv")
df.shape  # (42, 23)
 42 Public law; quoted from the official text, with a link to it 27 Oct 2026 26 Nov 2026 CSV JSON Releases

Asia

Asia →
Dataset Rows Terms Checked Due by Files
The Asia desk: official releases from China, Taiwan, South Korea and Japan on chips, export controls and AI, quoted One row per official release.
16 fields and code
  • id text
  • date text
  • country text
  • tag text
  • summary text
  • quote text
  • translation text
  • publisher text
  • source_title text
  • source_url text
  • kept_copy text
  • kept_date text
  • kept_sha256 text
  • page text
  • license text
  • notice text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/asia/asia.csv")
df.shape  # (25, 16)
 25 official government release; quoted, with a link to it; Open Government Data License, version 1.0 11 Nov 2026 11 Dec 2026 CSV JSON Releases

Dispatch

Dispatch →
Dataset Rows Terms Checked Due by Files
Dispatch event scores One row per event.
21 fields and code
  • edition text
  • from text
  • to text
  • event text
  • title text
  • organization text
  • date text
  • primary_link text
  • hn_points number
  • wiki_spike number
  • hf_upvotes text
  • yt_views number
  • hn_percentile number
  • wiki_percentile number
  • hf_percentile text
  • yt_percentile number
  • signals number
  • score number
  • qualified text
  • edition_page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/dispatch/scores.csv")
df.shape  # (5, 21)
 5 CC0 9 Oct 2026 on demand CSV JSON Releases

Records

Records →
Dataset Rows Terms Checked Due by Files
Records: where each series stands One row per series.
18 fields and code
  • series text
  • section text
  • newest_period text
  • newest_value number
  • unit text
  • preliminary text
  • standing text
  • since_period text
  • since_value number
  • previous_record_period text
  • previous_record_value number
  • readings number
  • first_period text
  • page text
  • source text
  • license text
  • retrieved text
  • series_page text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records.csv")
df.shape  # (18, 18)
 18 public domain; public filings; CC BY 4.0; cited as it asks 16 Oct 2026 10 Nov 2026 CSV JSON Releases
Records log: each new high and low One row per event.
16 fields and code
  • recorded text
  • snapshot number
  • series text
  • key text
  • record text
  • period text
  • value number
  • unit text
  • preliminary text
  • previous_record_period text
  • previous_record_value number
  • readings number
  • first_period text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/log.csv")
df.shape  # (8, 16)
 8 public domain; public filings 16 Oct 2026 10 Nov 2026 CSV JSON Releases
Records by state: each new high and low One row per event.
16 fields and code
  • recorded text
  • snapshot number
  • code text
  • state text
  • record text
  • period text
  • value number
  • unit text
  • preliminary text
  • previous_record_period text
  • previous_record_value number
  • readings number
  • first_period text
  • page text
  • source text
  • license text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/states-log.csv")
df.shape  # (38, 16)
 38 public domain 16 Oct 2026 20 Nov 2026 CSV JSON Releases
Records by state One row per state.
15 fields and code
  • code text
  • state text
  • newest_period text
  • newest_value number
  • unit text
  • standing text
  • compared_with text
  • compared_period text
  • compared_value number
  • readings number
  • first_period text
  • page text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/states.csv")
df.shape  # (51, 15)
 51 public domain 16 Oct 2026 20 Nov 2026 CSV JSON Releases
18 series files, one per series
Dataset Rows Terms Checked Due by Files
Records series: Capital spending of 9 AI spenders One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/capex.csv")
df.shape  # (10, 8)
 10 public filings 11 Oct 2026 10 Nov 2026 CSV JSON Releases
Records series: US chip output One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/chip-output.csv")
df.shape  # (116, 8)
 116 public domain 11 Oct 2026 15 Nov 2026 CSV JSON Releases
Records series: US chip plants’ use of capacity One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/chip-utilization.csv")
df.shape  # (116, 8)
 116 public domain 11 Oct 2026 15 Nov 2026 CSV JSON Releases
Records series: Unemployment of recent graduates One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/grads.csv")
df.shape  # (90, 8)
 90 cited as it asks 11 Oct 2026 10 Nov 2026 CSV JSON Releases
Records series: US hires rate One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/hires.csv")
df.shape  # (92, 8)
 92 public domain 11 Oct 2026 10 Nov 2026 CSV JSON Releases
Records series: US home power price One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/home-power.csv")
df.shape  # (127, 8)
 127 public domain 16 Oct 2026 20 Nov 2026 CSV JSON Releases
Records series: US imports of servers, twelve months One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/imports-servers.csv")
df.shape  # (93, 8)
 93 public domain 10 Oct 2026 14 Nov 2026 CSV JSON Releases
Records series: US imports of AI hardware, twelve months One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/imports.csv")
df.shape  # (93, 8)
 93 public domain 10 Oct 2026 14 Nov 2026 CSV JSON Releases
Records series: US layoffs rate One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/layoffs.csv")
df.shape  # (92, 8)
 92 public domain 11 Oct 2026 10 Nov 2026 CSV JSON Releases
Records series: US job openings rate One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/openings.csv")
df.shape  # (92, 8)
 92 public domain 11 Oct 2026 10 Nov 2026 CSV JSON Releases
Records series: US payroll jobs One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/payrolls.csv")
df.shape  # (93, 8)
 93 public domain 11 Oct 2026 10 Nov 2026 CSV JSON Releases
Records series: US job postings on Indeed One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/postings.csv")
df.shape  # (81, 8)
 81 CC BY 4.0 11 Oct 2026 10 Nov 2026 CSV JSON Releases
Records series: Producer prices: circuit boards and modules One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/ppi-boards.csv")
df.shape  # (273, 8)
 273 public domain 11 Oct 2026 15 Nov 2026 CSV JSON Releases
Records series: Producer prices: chips One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/ppi-chips.csv")
df.shape  # (715, 8)
 715 public domain 11 Oct 2026 15 Nov 2026 CSV JSON Releases
Records series: Producer prices: servers One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/ppi-servers.csv")
df.shape  # (261, 8)
 261 public domain 11 Oct 2026 15 Nov 2026 CSV JSON Releases
Records series: Producer prices: storage drives One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/ppi-storage.csv")
df.shape  # (405, 8)
 405 public domain 11 Oct 2026 15 Nov 2026 CSV JSON Releases
Records series: US data-center construction One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/spending.csv")
df.shape  # (152, 8)
 152 public domain 16 Oct 2026 11 Nov 2026 CSV JSON Releases
Records series: US unemployment rate One row per reading.
8 fields and code
  • period text
  • value number
  • unit text
  • preliminary text
  • series text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/records/unemployment.csv")
df.shape  # (92, 8)
 92 public domain 11 Oct 2026 10 Nov 2026 CSV JSON Releases

Grapher

Grapher →
Dataset Rows Terms Checked Due by Files
Grapher: every reading of every series One row per reading of a series.
9 fields and code
  • series text
  • label text
  • period text
  • value number
  • unit text
  • preliminary text
  • source text
  • license text
  • retrieved text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/series.csv")
df.shape  # (3906, 9)
 3,906 public domain; public filings; CC BY 4.0; cited as it asks; free reuse with credit; free to reuse with credit to UN Comtrade; Figures cited from public reports 20 Oct 2026 10 Nov 2026 CSV JSON Releases

Companies

Companies →
Dataset Rows Terms Checked Due by Files
Companies across every dataset One row per company.
29 fields and code
  • company text
  • slug text
  • sections text
  • capex_quarter text
  • capex_usd_m number
  • capex_trailing4_usd_m number
  • campuses_owned number
  • campuses_used number
  • owned_mw_now number
  • owned_mw_peak number
  • owned_h100e_now number
  • clusters number
  • cluster_h100e_existing number
  • cluster_h100e_planned number
  • cloud_regions_live number
  • cloud_regions_announced number
  • cloud_countries number
  • nuclear_agreements number
  • nuclear_mw_contract number
  • nuclear_mw_framework number
  • frontier_models number
  • frontier_days_at_top number
  • chip_plants number
  • chip_plants_planned number
  • taiwan_sales_month text
  • taiwan_sales_ntd_thousand number
  • url text
  • sources text
  • licenses text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/companies.csv")
df.shape  # (18, 29)
 18 CC BY 4.0; public filings; facts from public region lists; facts, quoted; CC BY-SA 4.0; OGDL 1.0 30 Oct 2026 6 Nov 2026 CSV JSON Releases

Explorer

Explorer →
Dataset Rows Terms Checked Due by Files
Explorer: AI campuses One row per campus.
14 fields and code
  • id text
  • name text
  • continent text
  • country text
  • owner text
  • power_now number
  • it_now number
  • h100e_now number
  • cost_now number
  • peak_power number
  • peak_date text
  • first_powered text
  • page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/explorer/campuses.csv")
df.shape  # (95, 14)
 95 CC BY 4.0 22 Oct 2026 10 Nov 2026 CSV JSON Releases
Explorer: countries One row per country.
16 fields and code
  • id text
  • name text
  • continent text
  • compute number
  • mw_now number
  • mw_end number
  • campuses number
  • supercomputers number
  • cloud_live number
  • cloud_announced number
  • providers number
  • fossil number
  • low_carbon number
  • intensity number
  • page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/explorer/countries.csv")
df.shape  # (54, 16)
 54 CC BY 4.0 11 Oct 2026 10 Nov 2026 CSV JSON Releases
Explorer: supercomputers One row per supercomputer.
10 fields and code
  • id text
  • name text
  • continent text
  • country text
  • owner text
  • certainty text
  • h100e number
  • date text
  • page text
  • source text
import pandas as pd

df = pd.read_csv("https://jesseahawkins.com/explorer/supercomputers.csv")
df.shape  # (453, 10)
 453 CC BY 4.0 11 Oct 2026 10 Nov 2026 CSV JSON Releases

Terms

—

Each file names its sources in its own columns, and the terms above are the words the file itself uses. Credit the source a file names when you reuse it. In the JSON copies, a column of plain numbers holds numbers, and an empty cell holds no value.