← Cockpit
INF_016predictionEnergypower-as-binding-constraint

Power generation is the 'innermost loop' and absolute limiting factor on AI expansion; any AI capability scaling ultimately routes through the speed at which new generation can be brought online.

Predictor: Elon Musk

Prior probability
85.0%
Current probability
57.5%
evolves via intake + LBP
Conviction
5/5
Signal quality
A
Resolution
in_progress
Window
2025-01-01 – 2030-10-31
Edges in / out
4 / 0
Tickers exposed
0

Prediction text

Power generation is the 'innermost loop' and absolute limiting factor on AI expansion; any AI capability scaling ultimately routes through the speed at which new generation can be brought online. | Colossus Phase 2 power contracts

Key catalyst: Colossus Phase 2 power contracts

Watch events: xAI / Tesla firm-power contracts; Colossus GW additions; Texas ERCOT capacity auctions

Resolution evidence

Status: in_progress

xAI Memphis deployed 35+ on-site gas turbines after grid interconnection delays; Morgan Stanley 49 GW US shortfall confirms binding-constraint thesis.

Predictor: Elon Musk

κ + Brier as of 2026-05-22
κ (discount)
0.688
Brier
0.0142
excellent
Hits / Misses
1 / 0
of 3 resolved
Hit rate
33.3%
Calibration plot (stated vs observed)

Evidence about this node from Elon Musk is multiplied by κ in /api/intake. Lower κ = less weight; floors at 0.10 (effectively silenced) and caps at 1.00 (full weight).

Reference class: energy_grid_rebuild_5y

Linked

Major grid expansion (>50GW) within 5y of announce

Base rate
50.0%
3/6 historical
Inside weight
Outside weight
no pull
inside 57.5% → blend 57.5% 0.0pp)

Tetlock-style outside view: at TRF=1 (just predicted), outside view dominates (w_in=0.3). At TRF=0 (deadline), inside view dominates (w_in=1.0). The blend regularizes overconfident inside views toward the historical base rate.

Probability over time

7 prob_history rows
0%25%50%75%100%prior 85%2026-04-302026-05-102026-05-24
intake v2milestone miss sweeplbp propagationreference class assignedlegacy v1prior_prob (analyst seed)current = 57.5%

Milestone chain

Pre-event signals (upstream prereqs + window checkpoints) → resolution event → downstream cascades. Status/dates update from linked nodes; re-derive nightly via scripts/ops/derive_milestones.py.
Leading chain: 2 fired ✓ · 1 overdue ⏱ · 7 pending
  1. 2026-01-15partialColossus 2 reaches 2 GW capacity with 555K GPUs
    How: xAI publicly announces or third party confirms that Colossus 2 has reached 2 gigawatts of total power capacity with ~555,000 NVIDIA Blackwell GPUs deployed
    Source: Introl — xAI Colossus Hits 2 GW: 555,000 GPUs, $18B (January 2026)conf 85%
    Notes: PARTIAL — xAI claimed 2GW; satellite imagery and Tom's Hardware reporting suggest only ~350MW cooling capacity is actually operational. Direct evidence that POWER is the binding constraint, not GPUs.
  2. 2026-02-13overdueQ1 window check-in (25%)
  3. 2026-02-15hitTom's Hardware exposes Colossus 2 power-cooling constraint
    How: Independent satellite imagery / journalism reveals that Colossus 2's cooling capacity (~350 MW) is the binding constraint, not GPU count
    Source: Tom's Hardware — xAI Colossus 2 nowhere near 1 GW, satellite imagery suggestsconf 95%
    Notes: HIT — Direct evidence supporting Musk's 'innermost loop' thesis. xAI has the GPUs but power/cooling is the rate-limiting step. Epoch AI projects 1 GW reality not until May 2026.
  4. 2026-04-01 → 2026-09-30pendingxAI Memphis natural gas power plant reaches commercial operation
    How: xAI's purpose-built natural gas plant adjacent to Memphis Colossus campus enters commercial operation per Tennessee Valley Authority filings
    Source: Interesting Engineering — Musk doubles down on Colossus supercomputerconf 65%
  5. 2026-06-01 → 2027-06-30pendingMajor hyperscaler (OpenAI / Anthropic / Google) publicly cites power as primary capacity constraint
    How: OpenAI, Anthropic, Google DeepMind, or Microsoft Azure publicly states in earnings call or press conference that AI scaling is gated by power generation, not GPU supply
    Source: Anticipated — quarterly earnings transcripts, AI infrastructure conferencesconf 85%
    Notes: Confirms thesis is consensus across labs, not just Musk's framing.
  6. 2026-04-01 → 2027-12-31pendingFirst multi-gigawatt PPA signed by frontier AI lab
    How: OpenAI, Anthropic, xAI, Microsoft, or Google announces a >2 GW power purchase agreement (nuclear, gas, or geothermal) with utility or independent power producer
    Source: Anticipated — utility 8-K filings, AI lab announcementsconf 75%
    Notes: Cascade — direct evidence that capital is rotating to power generation, the 'innermost loop'.
  7. 2027-03-29pendingQ2 window check-in (50%)
  8. 2028-05-10pendingQ3 window check-in (75%)
  9. 2029-03-31pendingScenario fires: AGI mid: Kurzweil 2029 path

No downstream cascades — this prediction is a leaf in the dependency graph.

What if this resolves?

Clamp this prediction TRUE or FALSE and run a counterfactual Gibbs sample. Surfaces the predictions whose marginals shift most under that assumption.
(live posterior: 57%)

Click a button to clamp this prediction and run a Gibbs sample. Returns the predictions whose marginals shift most. ~30s per run; ideal for stress-testing "if X resolves, what else moves?"

Evidence chain

Every probability update with full Bayesian provenance — chronological, latest first
LBP2026-05-24T02:00:02Z57.5%-12.3pp
Network propagation: 69.8% → 57.5%
4-iter LBP, residual 0.01000 · damping 0.5, w_intrinsic 0.5 · method lbp_v3 · run 806b02f8
intake_event_update2026-05-21T23:15:16Z69.8%+24.7pp
intake:7afeeb9a-f217-4dd2-b910-24ff14bdfc39 bayesian_v2 inside=0.698 blend=0.698 LLR=1.035 κ=0.64 no_blend
Raw metadata
{
  "trf": 0.7623443234575875,
  "kappa": 0.6429,
  "base_rate": null,
  "predictor": "Elon Musk",
  "total_llr": 1.6094379124341,
  "bayesian_v2": true,
  "prior_logit": -0.1983022027999053,
  "bayes_factor": "2.8:1 favoring",
  "blend_reason": "no reference_class linked",
  "inside_prior": 0.4505862710887746,
  "kappa_source": "predictor_table",
  "blend_applied": false,
  "contributions": [
    {
      "llr": 1.6094379124341,
      "kappa": 0.6429,
      "label": "Musk's 'innermost loop' thesis crystallizing - PJM reserve margin shortfall now imminent and explicit.",
      "adjusted_llr": 1.034707633903883
    }
  ],
  "evidence_kind": "intake_event_update",
  "inside_source": "history_v2",
  "inside_weight": 1,
  "outside_weight": 0,
  "posterior_prob": 0.6977076185949779,
  "evidence_origin": "daily_intake",
  "llm_suggestions": [
    {
      "polarity": "corroborates",
      "status_change": "unchanged",
      "evidence_strength": "strong",
      "delta_prob_suggestion": 0.06
    }
  ],
  "posterior_logit": 0.8364054311039779,
  "predictor_brier": 0.01,
  "evidence_doc_ids": [],
  "inside_posterior": 0.6977076185949779,
  "blended_posterior": 0.6977076185949779,
  "reference_class_id": null,
  "total_adjusted_llr": 1.034707633903883,
  "predictor_n_resolved": 2
}
LBP2026-05-17T02:00:01Z45.1%-2.5pp
Network propagation: 47.5% → 45.1%
5-iter LBP, residual 0.00689 · damping 0.5, w_intrinsic 0.5 · method lbp_v3 · run e607fa96
LBP2026-05-10T02:00:02Z47.5%-4.9pp
Network propagation: 52.4% → 47.5%
6-iter LBP, residual 0.00584 · damping 0.5, w_intrinsic 0.5 · method lbp_v3 · run e5c18d29
LBP2026-05-03T02:00:01Z52.4%-9.5pp
Network propagation: 61.9% → 52.4%
6-iter LBP, residual 0.00677 · damping 0.5, w_intrinsic 0.5 · method lbp_v3 · run 1a683ac9
metadata_milestone_miss_sweep2026-05-02T22:07:21Z61.9%-5.9pp
metadata_milestone_miss_sweep bayesian_v2 n=1 inside=0.619 blend=0.619 LLR=-0.261 κ=0.64 no_blend
Raw metadata
{
  "trf": 0.7712908523812041,
  "kappa": 0.6429,
  "base_rate": null,
  "predictor": "Elon Musk",
  "total_llr": -0.4054651081081644,
  "grace_days": 7,
  "bayesian_v2": true,
  "prior_logit": 0.7474588823823616,
  "bayes_factor": "1.3:1 against",
  "blend_reason": "no reference_class linked",
  "inside_prior": 0.6786247504453383,
  "kappa_source": "predictor_table",
  "n_milestones": 1,
  "blend_applied": false,
  "contributions": [
    {
      "llr": -0.4054651081081644,
      "kind": "quartile_checkpoint",
      "kappa": 0.6429,
      "label": "Q1 window check-in (25%)",
      "weight": 0.05,
      "strength": "weak",
      "confidence": null,
      "source_url": null,
      "adjusted_llr": -0.2606735180027389,
      "expected_date": "2026-02-13",
      "measurement_criterion": null
    }
  ],
  "evidence_kind": "metadata_milestone_miss_sweep",
  "inside_source": "history_v2",
  "inside_weight": 0.4600964033331571,
  "outside_weight": 0.5399035966668428,
  "posterior_prob": 0.6193488544940183,
  "posterior_logit": 0.48678536437962266,
  "predictor_brier": 0.01,
  "inside_posterior": 0.6193488544940183,
  "blended_posterior": 0.6193488544940183,
  "reference_class_id": null,
  "total_adjusted_llr": -0.2606735180027389,
  "predictor_n_resolved": 2
}
LBP2026-04-30T16:39:51Z67.9%-17.1pp
Network propagation: 85.0% → 67.9%
5-iter LBP, residual 0.00825 · damping 0.5, w_intrinsic 0.5 · method lbp_v2 · run 0c8a4ea3

Network propagation neighbors

Top edges sorted by latest LBP cross-impact
All propagation →

No propagation data yet. Run inference/.venv/bin/python scripts/ops/run_loopy_belief_propagation.py on the droplet, or wait for the Sunday 02:00 UTC weekly cron.

Prerequisites (4)

Predictions that must hit first
TypePredTitleDomainLag
correlateS_AGI_MID_2029AGI mid: Kurzweil 2029 pathagi_general_capability
correlateS_AGI_FAST_2027AGI fast: drop-in remote worker by 2027-09agi_general_capability
correlateS_AGI_SLOW_2031AGI slow: Schmidt/Hassabis 5-10 year pathagi_general_capability
correlateS_AGI_WINTER_2036PLUSAGI delayed: capability plateau or AI winteragi_general_capability

Dependents (0)

Predictions enabled by this
TypePredTitleDomainLag
No dependents

Expected milestones (1)

From Sheet 17 Monitoring Triggers
Expected byDescriptionStatus
2030-12-31[Energy/Regulation 2030-12] nnouncements; utility capex disclosures [INF_016] xAI / Tesla firm-power contracts; Colossus GW additions; Texas ERCOT capacity auctions [INF_013] Oklo / X-energy / Kairos Power deployments; FERC gas-pipeline approvals; hyperscaler PPA disclosurespending

Validations (1)

Resolution events
Observed atStatusByNotes
2026-04-29partialthesis_timeline_v1.0_importxAI Memphis deployed 35+ on-site gas turbines after grid interconnection delays; Morgan Stanley 49 GW US shortfall confirms binding-constraint thesis.

Raw metadata

From Thesis_Timeline_v1.0_FINAL workbook
{
  "nia": false,
  "mode": "FORECAST",
  "role": "Cited-CEO",
  "context": "Musk has repeatedly characterized power, not chips or talent, as the binding constraint. Demonstrated by xAI Memphis Colossus building on-site natural-gas turbines after grid delays.",
  "to_year": 2030,
  "conv_cues": "innermost loop; absolute limit; CEO FIRST_PERSON",
  "direction": "HAPPEN",
  "from_year": 2025,
  "timeframe": "ongoing",
  "conv_level": "HIGH",
  "milestones": [
    {
      "kind": "llm_pre_event",
      "label": "Colossus 2 reaches 2 GW capacity with 555K GPUs",
      "notes": "PARTIAL — xAI claimed 2GW; satellite imagery and Tom's Hardware reporting suggest only ~350MW cooling capacity is actually operational. Direct evidence that POWER is the binding constraint, not GPUs.",
      "source": "Introl — xAI Colossus Hits 2 GW: 555,000 GPUs, $18B (January 2026)",
      "status": "partial",
      "weight": 0.4,
      "ordinal": -10,
      "source_id": null,
      "confidence": 0.85,
      "source_url": "https://introl.com/blog/xai-colossus-2-gigawatt-expansion-555k-gpus-january-2026",
      "expected_date": "2026-01-31",
      "observed_date": "2026-01-15",
      "research_origin": "deep_research",
      "measurement_criterion": "xAI publicly announces or third party confirms that Colossus 2 has reached 2 gigawatts of total power capacity with ~555,000 NVIDIA Blackwell GPUs deployed"
    },
    {
      "kind": "quartile_checkpoint",
      "label": "Q1 window check-in (25%)",
      "status": "overdue",
      "weight": 0.05,
      "ordinal": -9,
      "source_id": null,
      "expected_date": "2026-02-13",
      "observed_date": null,
      "miss_emitted_at": "2026-05-02T22:07:21.384228+00:00",
      "miss_emitted_by": "metadata_milestone_sweep"
    },
    {
      "kind": "llm_pre_event",
      "label": "Tom's Hardware exposes Colossus 2 power-cooling constraint",
      "notes": "HIT — Direct evidence supporting Musk's 'innermost loop' thesis. xAI has the GPUs but power/cooling is the rate-limiting step. Epoch AI projects 1 GW reality not until May 2026.",
      "source": "Tom's Hardware — xAI Colossus 2 nowhere near 1 GW, satellite imagery suggests",
      "status": "hit",
      "weight": 0.4,
      "ordinal": -8,
      "source_id": null,
      "confidence": 0.95,
      "source_url": "https://www.tomshardware.com/tech-industry/artificial-intelligence/elon-musks-xai-colossus-2-is-nowhere-near-1-gigawatt-capacity-satellite-imagery-suggests-despite-claims-site-only-has-350-megawatts-of-cooling-capacity",
      "expected_date": "2026-02-15",
      "observed_date": "2026-02-15",
      "research_origin": "deep_research",
      "measurement_criterion": "Independent satellite imagery / journalism reveals that Colossus 2's cooling capacity (~350 MW) is the binding constraint, not GPU count"
    },
    {
      "kind": "llm_pre_event",
      "label": "xAI Memphis natural gas power plant reaches commercial operation",
      "source": "Interesting Engineering — Musk doubles down on Colossus supercomputer",
      "status": "pending",
      "weight": 0.4,
      "ordinal": -7,
      "source_id": null,
      "confidence": 0.65,
      "source_url": "https://interestingengineering.com/ai-robotics/elon-musk-doubles-down-on-colossus-supercomputer",
      "expected_date": "2026-07-01",
      "research_origin": "deep_research",
      "expected_date_range": {
        "to": "2026-09-30",
        "from": "2026-04-01"
      },
      "measurement_criterion": "xAI's purpose-built natural gas plant adjacent to Memphis Colossus campus enters commercial operation per Tennessee Valley Authority filings"
    },
    {
      "kind": "llm_pre_event",
      "label": "Major hyperscaler (OpenAI / Anthropic / Google) publicly cites power as primary capacity constraint",
      "notes": "Confirms thesis is consensus across labs, not just Musk's framing.",
      "source": "Anticipated — quarterly earnings transcripts, AI infrastructure conferences",
      "status": "pending",
  
... (truncated)