Skip to content
Slow Variables

Ask the data

Answers come from the same store as the site; every number is checked against the record it cites.

Memos · 2026-09-03 to 2026-09-10

Week to 10 September 2026

Deterministic digest (no ANTHROPIC_API_KEY); approved by merge.

Diffusion lens
As of 2026-09-03: methods mixed, products consistent with normal, early adoption consistent with normal, adaptation consistent with normal, return arrow emerging.
Capture lens
As of 2026-09-10: compute & physical concentrating, model consistent with normal, training input faster than normal, serving & orchestration unmeasured, deployment & application consistent with normal, adopters consistent with normal, labour & consumers consistent with normal.

Thesis snapshot: Capture thesis 'consumers keep most of the surplus' HOLDS holds · Capture thesis 'rents migrate up the stack' SUPPORTED does not hold · Invention-side WARNING (bottleneck #86 under test) untestable · Normal-technology thesis FALSIFIED does not hold · Normal-technology thesis STRENGTHENED does not hold

As of 2026-09-03: methods mixed, products consistent with normal, early adoption consistent with normal, adaptation consistent with normal, return arrow emerging. As of 2026-09-10: compute & physical concentrating, model consistent with normal, training input faster than normal, serving & orchestration unmeasured, deployment & application consistent with normal, adopters consistent with normal, labour & consumers consistent with normal.

What changed

  • metr_horizon_50: unmeasured → emerging, confidence 80 ind:metr_hor, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:a790bbf0.
  • metr_horizon_80: unmeasured → emerging, confidence 70 ind:metr_hor, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:1bd721a4.
  • horizon_ratio_80_50: unmeasured → consistent with normal, confidence 70 ind:horizon_, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:aaf7234c.
  • semis_rent_concentration: unmeasured → concentrating, confidence 85 ind:semis_re, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:3e75261b.
  • cloud_rpo_backlog: unmeasured → concentrating, confidence 80 ind:cloud_rp, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:1af173b0.
  • metr_horizon_50: emerging → faster than normal, confidence 80 ind:metr_hor, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:30eed8a5.
  • metr_horizon_80: emerging → faster than normal, confidence 70 ind:metr_hor, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:7934728c.
  • bls_tfp_private_nonfarm: unmeasured → consistent with normal, confidence 75 ind:bls_tfp_, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:87a646b2.
  • consumer_surplus_wta: unmeasured → dispersing, confidence 65 ind:consumer, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:97faa342.
  • bls_labor_productivity_yoy: unmeasured → consistent with normal, confidence 80 ind:bls_labo, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:3c9ce7da.
  • dev_rct_uplift: unmeasured → consistent with normal, confidence 70 ind:dev_rct_, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:a29f4505.
  • enterprise_pilot_to_production: unmeasured → consistent with normal, confidence 55 ind:enterpri, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:404c8c6c.
  • bbd_work_hours_assisted: unmeasured → consistent with normal, confidence 80 ind:bbd_work, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:e786ebb3.
  • btos_firm_use: unmeasured → consistent with normal, confidence 70 ind:btos_fir, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:61b8d268.
  • canaries_entry_level_gap: unmeasured → emerging, confidence 70 ind:canaries, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:19f8e3c4.
  • revelio_exposed_vs_unexposed_growth: unmeasured → emerging, confidence 65 ind:revelio_, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:c9553f06.
  • cait_high_exposure_claims: unmeasured → consistent with normal, confidence 70 ind:cait_hig, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:c674167a.
  • new_grad_unemployment: unmeasured → consistent with normal, confidence 75 ind:new_grad, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:047e2d4c.
  • cross_tracker_concordance: unmeasured → consistent with normal, confidence 70 ind:cross_tr, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:162319d3.
  • aei_augmentation_share: unmeasured → consistent with normal, confidence 60 ind:aei_augm, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:c18c9e4a.
  • labor_share_nonfarm: unmeasured → emerging, confidence 65 ind:labor_sh, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:683517b9.
  • cloud_segment_margins: unmeasured → stable, confidence 75 ind:cloud_se, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:5aef343d.
  • margin_stack_semis_share: unmeasured → concentrating, confidence 70 ind:margin_s, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:eacb8a9a.
  • circular_financing_scale: unmeasured → concentrating, confidence 75 ind:circular, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:0bf1bcda.
  • ramp_paid_ai_adoption: unmeasured → emerging, confidence 60 ind:ramp_pai, by evaluate on 2026-09-10; the reason is in the changelog event:4acc0bfa.
  • openai_research_intern_sep_2026: unmeasured → emerging, confidence 30 ind:openai_r, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:155c4776.
  • openai_automated_researcher_mar_2028: unmeasured → not yet testable, confidence 30 ind:openai_a, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:78e2ea21.
  • amodei_swe_end_to_end_6_12mo: unmeasured → behind, confidence 55 ind:amodei_s, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:b3b7c06c.
  • amodei_country_of_geniuses_1_2y: unmeasured → not yet testable, confidence 30 ind:amodei_c, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:9622fef3.
  • amodei_trillions_before_2030: unmeasured → emerging, confidence 35 ind:amodei_t, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:d527928f.
  • anthropic_10x_per_year: unmeasured → ahead, confidence 50 ind:anthropi, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:6e76cc5e.
  • ai2027_horizon_doubling_4mo: unmeasured → on track, confidence 60 ind:ai2027_h, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:f25d784b.
  • ai2027_80pct_years_by_mar_2027: unmeasured → behind, confidence 55 ind:ai2027_8, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:3a46e6d2.
  • ai2027_superhuman_coder_2027: unmeasured → not yet testable, confidence 30 ind:ai2027_s, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:8d28bfab.
  • cahn_600b_question: unmeasured → emerging, confidence 40 ind:cahn_600, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:e7e407dd.
  • cahn_1_5t_question: unmeasured → emerging, confidence 40 ind:cahn_1_5, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:87bedeea.
  • bain_2t_revenue_2030: unmeasured → not yet testable, confidence 30 ind:bain_2t_, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:a779afb3.
  • covello_too_much_spend: unmeasured → emerging, confidence 40 ind:covello_, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:73f5baf7.
  • tunguz_12x_infra_per_revenue: unmeasured → emerging, confidence 35 ind:tunguz_1, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:25bcd5ba.
  • korinek_modest_scenario: unmeasured → not yet testable, confidence 30 ind:korinek_, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:05865571.
  • korinek_substantial_scenario: unmeasured → not yet testable, confidence 30 ind:korinek_, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:6b0913e9.
  • korinek_extreme_scenario: unmeasured → not yet testable, confidence 30 ind:korinek_, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:fbdb2cf8.
  • korinek_divergence_after_2027: unmeasured → not yet testable, confidence 30 ind:korinek_, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:61cfa81c.
  • nk_adoption_decades: unmeasured → on track, confidence 70 ind:nk_adopt, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:a9768aa5.
  • nk_safety_speed_limits: unmeasured → not yet testable, confidence 30 ind:nk_safet, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:88a4cdad.
  • nk_rsi_external_limits: unmeasured → emerging, confidence 40 ind:nk_rsi_e, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:98805453.
  • nk_benchmarks_false_summits: unmeasured → on track, confidence 65 ind:nk_bench, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:255a3f1b.
  • nk_agi_not_milestone: unmeasured → not yet testable, confidence 30 ind:nk_agi_n, by claude-initial-seed on 2026-09-10; the reason is in the changelog event:a993e859.
  • rsi_agent_workdays_per_human: unmeasured → emerging, confidence 30 ind:rsi_agen, by evaluate on 2026-09-10; the reason is in the changelog event:4838e15c.
  • rsi_intervention_rate_4_8h: unmeasured → emerging, confidence 30 ind:rsi_inte, by evaluate on 2026-09-10; the reason is in the changelog event:94e0a387.
  • enterprise_spend_by_layer: unmeasured → emerging, confidence 55 ind:enterpri, by evaluate on 2026-09-10; the reason is in the changelog event:095d3d96.
  • semis_hhi: unmeasured → unclear, confidence 60 ind:semis_hh, by evaluate on 2026-09-10; the reason is in the changelog event:ae11b6d8.
  • cloud_hhi: unmeasured → dispersing, confidence 60 ind:cloud_hh, by evaluate on 2026-09-10; the reason is in the changelog event:9afa19bb.
  • saas_multiples: unmeasured → emerging, confidence 50 ind:saas_mul, by evaluate on 2026-09-10; the reason is in the changelog event:ea9864fc.
  • model_price_per_horizon_hour: unmeasured → not yet measurable, confidence 40 ind:model_pr, by evaluate on 2026-09-10; the reason is in the changelog event:2c653b55.
  • frontier_capital_concentration: unmeasured → stable, confidence 50 ind:frontier, by evaluate on 2026-09-10; the reason is in the changelog event:3ad1849a.
  • vertical_vs_horizontal_apps: unmeasured → dispersing, confidence 45 ind:vertical, by evaluate on 2026-09-10; the reason is in the changelog event:da628968.
  • lab_valuation_to_run_rate: unmeasured → emerging, confidence 45 ind:lab_valu, by evaluate on 2026-09-10; the reason is in the changelog event:5029df3d.
  • open_vs_closed_gap: unmeasured → dispersing, confidence 55 ind:open_vs_, by evaluate on 2026-09-10; the reason is in the changelog event:6eeb9a97.
  • custom_silicon_share: unmeasured → emerging, confidence 50 ind:custom_s, by evaluate on 2026-09-10; the reason is in the changelog event:de691085.
  • epoch_training_compute_growth: unmeasured → emerging, confidence 60 ind:epoch_tr, by evaluate on 2026-09-10; the reason is in the changelog event:fc17d3a0.
  • epoch_training_power_growth: unmeasured → emerging, confidence 50 ind:epoch_tr, by evaluate on 2026-09-10; the reason is in the changelog event:04857236.
  • perf_per_dollar_growth: unmeasured → consistent with normal, confidence 40 ind:perf_per, by evaluate on 2026-09-10; the reason is in the changelog event:6f770f1e.
  • epoch_inference_price_fixed_capability: unmeasured → emerging, confidence 45 ind:epoch_in, by evaluate on 2026-09-10; the reason is in the changelog event:f5c95d74.
  • arc_agi_frontier: unmeasured → emerging, confidence 55 ind:arc_agi_, by evaluate on 2026-09-10; the reason is in the changelog event:760e7d97.
  • safety_brake_events: unmeasured → emerging, confidence 40 ind:safety_b, by evaluate on 2026-09-10; the reason is in the changelog event:7a7dea71.
  • epoch_data_exhaustion_year: unmeasured → emerging, confidence 40 ind:epoch_da, by evaluate on 2026-09-10; the reason is in the changelog event:2a48514c.
  • fda_ai_devices_cumulative: unmeasured → consistent with normal, confidence 70 ind:fda_ai_d, by evaluate on 2026-09-10; the reason is in the changelog event:968ce880.
  • state_ai_bills_introduced: unmeasured → emerging, confidence 55 ind:state_ai, by evaluate on 2026-09-10; the reason is in the changelog event:cadcabc7.
  • rl_environment_vendor_count: unmeasured → emerging, confidence 35 ind:rl_envir, by evaluate on 2026-09-10; the reason is in the changelog event:d7c0a996.
  • capex_to_revenue_stack: unmeasured → dispersing, confidence 40 ind:capex_to, by evaluate on 2026-09-10; the reason is in the changelog event:9d4f4e62.
  • ord_half_life_misfit: unmeasured → consistent with normal, confidence 60 ind:ord_half, by evaluate on 2026-09-10; the reason is in the changelog event:7331962a.
  • startup_share_of_app_revenue: unmeasured → emerging, confidence 50 ind:startup_, by evaluate on 2026-09-10; the reason is in the changelog event:38272114.
  • continual_learning_level: unmeasured → emerging, confidence 45 ind:continua, by evaluate on 2026-09-10; the reason is in the changelog event:0da1c10d.
  • lab_run_rates: unmeasured → concentrating, confidence 55 ind:lab_run_, by evaluate on 2026-09-10; the reason is in the changelog event:fb7eb16b.
  • lab_recoupment_ratio: unmeasured → emerging, confidence 50 ind:lab_reco, by evaluate on 2026-09-10; the reason is in the changelog event:6508557d.
  • fda_first_llm_device: unmeasured → consistent with normal, confidence 60 ind:fda_firs, by evaluate on 2026-09-10; the reason is in the changelog event:ec4184a8.
  • canaries_exposed_employment_yoy: unmeasured → consistent with normal, confidence 80 ind:canaries, by evaluate on 2026-09-10; the reason is in the changelog event:61072ea8.
  • call_center_uplift: unmeasured → consistent with normal, confidence 75 ind:call_cen, by evaluate on 2026-09-10; the reason is in the changelog event:60083764.
  • exec_reported_impact: unmeasured → consistent with normal, confidence 70 ind:exec_rep, by evaluate on 2026-09-10; the reason is in the changelog event:2db7af87.
  • harvey_lab_frontier_completion: unmeasured → emerging, confidence 40 ind:harvey_l, by evaluate on 2026-09-10; the reason is in the changelog event:26f5e7c4.
  • pass_hat_k_tau_bench: unmeasured → emerging, confidence 55 ind:pass_hat, by evaluate on 2026-09-10; the reason is in the changelog event:d67cec81.
  • surprise_index: unmeasured → emerging, confidence 50 ind:surprise, by evaluate on 2026-09-10; the reason is in the changelog event:386e02ff.
  • openai_work_share: unmeasured → emerging, confidence 65 ind:openai_w, by evaluate on 2026-09-10; the reason is in the changelog event:7a642e59.
  • exposure_vs_usage_gap: unmeasured → consistent with normal, confidence 60 ind:exposure, by evaluate on 2026-09-10; the reason is in the changelog event:2485c762.
  • exposed_low_adaptive_workers: unmeasured → emerging, confidence 45 ind:exposed_, by evaluate on 2026-09-10; the reason is in the changelog event:715f5567.
  • humlum_precise_null: unmeasured → consistent with normal, confidence 80 ind:humlum_p, by evaluate on 2026-09-10; the reason is in the changelog event:56ad1fba.
  • upwork_freelancer_effects: unmeasured → consistent with normal, confidence 70 ind:upwork_f, by evaluate on 2026-09-10; the reason is in the changelog event:a72e885f.
  • pwc_ai_skill_wage_premium: unmeasured → emerging, confidence 50 ind:pwc_ai_s, by evaluate on 2026-09-10; the reason is in the changelog event:95020797.
  • expert_data_market_run_rate: unmeasured → faster than normal, confidence 70 ind:expert_d, by evaluate on 2026-09-10; the reason is in the changelog event:2410fef4.
  • internal_external_deployment_gap_months: unmeasured → consistent with normal, confidence 45 ind:internal, by evaluate on 2026-09-10; the reason is in the changelog event:33785c35.
  • waymo_weekly_paid_rides: unmeasured → consistent with normal, confidence 65 ind:waymo_we, by evaluate on 2026-09-10; the reason is in the changelog event:fca7d874.

New evidence

Watchlist

Thesis monitor

  • Normal-technology thesis FALSIFIED: does not hold
  • Normal-technology thesis STRENGTHENED: does not hold
  • Invention-side WARNING (bottleneck #86 under test): untestable
  • Capture thesis 'rents migrate up the stack' SUPPORTED: does not hold
  • Capture thesis 'consumers keep most of the surplus' HOLDS: holds

Crosswalk

  • methods ⇄ model moved in opposite directions: metr_horizon_50, metr_horizon_80, horizon_ratio_80_50, metr_horizon_50, metr_horizon_80, epoch_training_compute_growth, epoch_training_power_growth, perf_per_dollar_growth, epoch_inference_price_fixed_capability, arc_agi_frontier, epoch_data_exhaustion_year, ord_half_life_misfit, surprise_index, expert_data_market_run_rate versus metr_horizon_50, metr_horizon_80, horizon_ratio_80_50, metr_horizon_50, metr_horizon_80, rsi_agent_workdays_per_human, rsi_intervention_rate_4_8h, model_price_per_horizon_hour, frontier_capital_concentration, lab_valuation_to_run_rate, open_vs_closed_gap, epoch_training_compute_growth, epoch_inference_price_fixed_capability, arc_agi_frontier, safety_brake_events, ord_half_life_misfit, lab_run_rates, lab_recoupment_ratio, pass_hat_k_tau_bench, surprise_index, internal_external_deployment_gap_months
  • methods ⇄ compute_physical moved in opposite directions: metr_horizon_50, metr_horizon_80, horizon_ratio_80_50, metr_horizon_50, metr_horizon_80, epoch_training_compute_growth, epoch_training_power_growth, perf_per_dollar_growth, epoch_inference_price_fixed_capability, arc_agi_frontier, epoch_data_exhaustion_year, ord_half_life_misfit, surprise_index, expert_data_market_run_rate versus semis_rent_concentration, cloud_rpo_backlog, cloud_segment_margins, margin_stack_semis_share, circular_financing_scale, semis_hhi, cloud_hhi, custom_silicon_share, epoch_training_power_growth, perf_per_dollar_growth, capex_to_revenue_stack

Lens sentences

Diffusion: As of 2026-09-03: methods mixed, products consistent with normal, early adoption consistent with normal, adaptation consistent with normal, return arrow emerging.

Capture: As of 2026-09-10: compute & physical concentrating, model consistent with normal, training input faster than normal, serving & orchestration unmeasured, deployment & application consistent with normal, adopters consistent with normal, labour & consumers consistent with normal.