INF_012predictionEnergyAI-power-additions

AI data-center power consumption will grow by approximately 126 GW annually through 2028 — comparable to Canada's total annual power demand, and representing nearly 20% of projected global power growth through 2030.

Predictor: Morgan Stanley

Prior probability

60.0%

Current probability

46.0%

evolves via intake + LBP

Conviction

5/5

Signal quality

Resolution

pending

Window

2026-01-01 – 2028-11-30

Edges in / out

4 / 0

Tickers exposed

Prediction text

AI data-center power consumption will grow by approximately 126 GW annually through 2028 — comparable to Canada's total annual power demand, and representing nearly 20% of projected global power growth through 2030. | Transformer supply recovery; firm-power PPA velocity

Key catalyst: Transformer supply recovery; firm-power PPA velocity

Watch events: IEA annual electricity outlook; FERC ISO load-growth filings; transformer supply chain recovery

Resolution evidence

Status: pending

IEA 'Electricity 2026' projects DC electricity demand to double by 2030; consistent with Morgan Stanley 126 GW/yr.

Predictor: Morgan Stanley

κ + Brier as of 2026-05-22

Full calibration →

κ (discount)

0.633

Brier

0.0442

excellent

Hits / Misses

1 / 0

of 2 resolved

Hit rate

50.0%

Calibration plot (stated vs observed)

Evidence about this node from Morgan Stanley 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 via embedding similarity 0.655

All classes →

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

Base rate

50.0%

3/6 historical

Inside weight

0.399

TRF=0.86

Outside weight

0.601

pulling toward base rate

inside 40.1% → blend 46.0% (Δ 5.9pp)

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

intake v2milestone miss sweeplbp propagationreference class assignedlegacy v1prior_prob (analyst seed)current = 46.0%

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: 5 fired ✓ · 9 pending

2026-01-15hitMorgan Stanley publishes 'Powering AI' report with 126 GW/yr forecast
How: Morgan Stanley Research publishes 'Powering AI' (or equivalent) report quantifying AI DC power growth at >=100 GW/yr through 2028
Source: https://www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026conf 99%
Notes: HIT — Morgan Stanley insights article confirms 126 GW/yr AI DC growth equivalent to Canada's annual demand.
2026-01-15hitMS forecasts US DC demand reaches 74 GW by 2028
How: Morgan Stanley research publication forecasts US DC demand >=70 GW by 2028 with explicit power-shortfall warning
Source: https://www.investing.com/news/stock-market-news/morgan-stanley-sees-up-to-20-shortage-of-us-power-for-data-centers-through-2028-4351251conf 97%
2026-02-28hitMorgan Stanley Powering AI report reaffirms 126 GW/yr through 2028
How: Morgan Stanley publishes 'Powering AI' update reaffirming ~126 GW annual data-center power additions through 2028
Source: https://www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026conf 95%
Notes: HIT — MS reaffirmed 126 GW/yr through 2028; framed as 1/5 of global power growth
2026-02-28hitMorgan Stanley quantifies US shortfall — 49 GW by 2028
How: Morgan Stanley Research forecasts US data-center demand 74 GW by 2028 with ~49 GW shortfall in available power access
Source: https://www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026conf 95%
2026-07-23pendingQ1 window check-in (25%)
2026-01-15hit$1T+ tech capex committed to AI infrastructure 2025-2026
How: Aggregate AI infrastructure capex announced by Big Tech (MSFT/GOOG/AMZN/META) for 2025-2026 exceeds $1 trillion per Morgan Stanley tracker
Source: https://www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026conf 95%
2027-02-12pendingQ2 window check-in (50%)
2026-06-01 → 2027-12-31pendingFirst grid stress event causes major data-center curtailment
How: Any US ISO (PJM, ERCOT, MISO) issues mandatory curtailment to large data center load during peak event with public reporting
Source: https://nzero.com/blog/u-s-power-demand-hits-new-highs-driven-by-data-centers-ai-and-grid-constraints/conf 70%
2026-09-01 → 2027-09-30pendingIEA annual outlook updates AI-DC share of global power growth ≥20%
How: IEA Electricity 2026/2027 report attributes ≥20% of incremental global electricity demand growth through 2030 to AI data centers
Source: https://www.goldmansachs.com/insights/articles/ai-to-drive-165-increase-in-data-center-power-demand-by-2030conf 85%
2026-09-01 → 2028-06-30pendingUS utility capex commitments cross $1.4T cumulative for AI grid build
How: US investor-owned utilities collectively commit ≥$1.4T capex in 2025-2030 IRPs explicitly tied to AI/DC load
Source: https://tech-insider.org/us-utility-1-4-trillion-ai-data-center-energy-2026/conf 75%
2026-06-01 → 2028-11-30pendingBehind-the-meter natural gas + nuclear PPAs reach 50 GW signed
How: Cumulative announced behind-the-meter natural gas + nuclear power deals for AI DCs reach 50 GW per BloombergNEF or S&P
Source: https://www.morganstanley.com/ideas/ai-energy-demand-infrastructureconf 60%
2027-09-03pendingQ3 window check-in (75%)
2027-12-31pendingScenario fires: 50GW dedicated AI/data center grid by Dec 2027
2027-01-01 → 2028-12-31pendingUS DC power shortfall reaches 44 GW (worst case)
How: EIA, NERC, or Morgan Stanley confirms US DC power shortfall >=40 GW with delayed grid interconnection queue
Source: https://mlq.ai/news/morgan-stanley-warns-of-looming-45-gigawatt-us-power-shortage-driven-by-ai-data-center-boom/conf 65%
2028-03-25pendingAI data-center power consumption will grow by approximately 126 GW annually through 2028 — comparable to Canada's total annual power demand,
2028-11-30pending126 GW/yr DC demand growth realized 2026-2028 cumulatively
How: Cumulative 2026-2028 AI DC commissioned capacity reaches >=300 GW (3 years × ~100 GW realized vs 126 GW forecast)
Source: https://www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026conf 45%
Notes: Cascade — Forecast vs realization gap likely 20-40% due to grid bottlenecks documented by Morgan Stanley.

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: 46%)

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

milestone miss2026-05-31T22:09:40Z46.0%-4.9pp

50.9%(inside prior)→40.1%(Bayes, κ=0.63)→46.0%(blend w_in=0.40 × 50% energy_grid_rebuild_5y)

1 milestone missed · 30d grace

Show missed milestones

LLR -0.693 · moderate · expected 2026-04-30
Morgan Stanley publishes 'Powering AI' report with 126 GW/yr forecast

LBP2026-05-24T02:00:02Z50.9%-2.6pp

Network propagation: 53.6% → 50.9%

4-iter LBP, residual 0.01000 · damping 0.5, w_intrinsic 0.5 · method lbp_v3 · run 806b02f8

intake_event_update2026-05-21T23:15:16Z53.6%+4.6pp

intake:7afeeb9a-f217-4dd2-b910-24ff14bdfc39 bayesian_v2 inside=0.590 blend=0.536 LLR=0.404 κ=0.58 w_in=0.39 energy_grid_rebuild_5y

Raw metadata

{
  "trf": 0.8675103990988756,
  "kappa": 0.5833,
  "base_rate": 0.5,
  "predictor": "Morgan Stanley",
  "total_llr": 0.6931471805599453,
  "bayesian_v2": true,
  "prior_logit": -0.04005452534874247,
  "bayes_factor": "1.5:1 favoring",
  "blend_reason": "blend 39% inside / 61% outside (TRF=0.868, base_rate=0.500 from energy_grid_rebuild_5y)",
  "inside_prior": 0.489987707241362,
  "kappa_source": "predictor_table",
  "blend_applied": true,
  "contributions": [
    {
      "llr": 0.6931471805599453,
      "kappa": 0.5833,
      "label": "PJM alone projecting 30 GW additional data center load by 2030.",
      "adjusted_llr": 0.4043127504206161
    }
  ],
  "evidence_kind": "intake_event_update",
  "inside_source": "history_v2",
  "inside_weight": 0.3927427206307871,
  "outside_weight": 0.6072572793692129,
  "posterior_prob": 0.5357040687653802,
  "evidence_origin": "daily_intake",
  "llm_suggestions": [
    {
      "polarity": "corroborates",
      "status_change": "unchanged",
      "evidence_strength": "moderate",
      "delta_prob_suggestion": 0.04
    }
  ],
  "posterior_logit": 0.3642582250718736,
  "predictor_brier": 0.01,
  "evidence_doc_ids": [],
  "inside_posterior": 0.5900708380323294,
  "blended_posterior": 0.5357040687653802,
  "reference_class_id": "energy_grid_rebuild_5y",
  "total_adjusted_llr": 0.4043127504206161,
  "predictor_n_resolved": 1
}

LBP2026-05-03T02:00:01Z49.0%-1.0pp

Network propagation: 50.0% → 49.0%

6-iter LBP, residual 0.00677 · damping 0.5, w_intrinsic 0.5 · method lbp_v3 · run 1a683ac9

milestone miss2026-05-01T22:09:22Z50.0%-3.8pp

60.0%(inside prior)→50.0%(Bayes, κ=0.58)→50.0%(blend w_in=0.38 × 50% energy_grid_rebuild_5y)

1 milestone missed · 30d grace

Show missed milestones

LLR -0.693 · moderate · expected 2026-01-31
MS forecasts US DC demand reaches 74 GW by 2028

legacy v12026-04-30T16:13:50Z53.8%+0.0pp

reference_class_assigned bayesian_v2 inside=0.600 blend=0.538 w_in=0.38 energy_grid_rebuild_5y

legacy v12026-04-30T01:56:50Z53.8%-6.2pp

reference_class_assigned bayesian_v2 inside=0.600 blend=0.538 w_in=0.38 energy_grid_rebuild_5y

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.

Ticker exposure

27 ticker(s) linked

Beneficiaries (21)

FSLR ARGAN WULF APLD IREN EQIX CRWV FLNC NBIS IRM META MSFT ETN ORCL SFTBY STX AAPL AMZN AMT GOOGL HUBB

Prerequisites (4)

Predictions that must hit first

Type	Pred	Title	Domain	Lag
correlate	S_COMPUTE_1GW_2027	Compute: 1GW operational by Jun 2027	compute_scale	—
correlate	S_GRID_50GW_2027	50GW dedicated AI/data center grid by Dec 2027	energy_grid_expansion	—
correlate	S_COMPUTE_10GW_2028	Compute: 10GW total by Dec 2028	compute_scale	—
correlate	S_COMPUTE_100GW_2030	Compute: 100GW national-scale by Dec 2030	compute_scale	—

Dependents (0)

Predictions enabled by this

Type	Pred	Title	Domain	Lag
No dependents

Expected milestones (7)

From Sheet 17 Monitoring Triggers

Expected by	Description	Status
2026-01-31	MS forecasts US DC demand reaches 74 GW by 2028	missed
2026-04-30	Morgan Stanley publishes 'Powering AI' report with 126 GW/yr forecast	missed
2026-12-31	$1T+ tech capex committed to AI infrastructure 2025-2026	pending
2028-11-30	Behind-the-meter natural gas + nuclear PPAs reach 50 GW signed	pending
2028-11-30	126 GW/yr DC demand growth realized 2026-2028 cumulatively	pending
2028-12-31	[Energy/Regulation 2028-12] e Mile Island, Palisades); interconnect [INF_012] IEA annual electricity outlook; FERC ISO load-growth filings; transformer supply chain recovery [INF_055] Polaris Q-factor disclosures; Orion groundbreaking; Helion first grid electricity	pending
2028-12-31	US DC power shortfall reaches 44 GW (worst case)	pending

Linked documents (10)

Auto-generated by cosine similarity from Polymarket / Manifold / EDGAR / GDELT

Sim	Source	Title	Market prob	Polarity	Reviewed	Published
0.734	manifold	Will data centers consume more than 10% of US electricity usage by EOY 2028?	65%	mentions	pending	2026-05-28
0.652	manifold	Will MISO Indiana Hub 2028 day-ahead annual average exceed US$60/MWh?	29%	mentions	pending	2026-05-04
0.645	manifold	Will MISO Indiana Hub 2027 day-ahead annual average exceed US$55/MWh?	33%	mentions	pending	2026-05-04
0.637	manifold	Will Nord Pool 2028 system price annual average exceed €45/MWh?	52%	mentions	pending	2026-05-04
0.633	manifold	Will Vic NEM 2028 annual average wholesale price exceed A$115/MWh?	32%	mentions	pending	2026-05-04
0.622	manifold	Will WA SWIS 2028 annual average wholesale price exceed A$150/MWh?	28%	mentions	pending	2026-05-04
0.622	manifold	Will Nord Pool 2027 system price annual average exceed €45/MWh?	50%	mentions	pending	2026-05-04
0.620	manifold	Will Vic NEM 2027 annual average wholesale price exceed A$105/MWh?	35%	mentions	pending	2026-05-04
0.608	manifold	Will WA SWIS 2027 annual average wholesale price exceed A$135/MWh?	25%	mentions	pending	2026-05-04
0.537	edgar_8k	PLUG POWER INC (PLUG) (CIK 0001093691)	—	mentions	pending	2026-05-11

Raw metadata

From Thesis_Timeline_v1.0_FINAL workbook

{
  "nia": false,
  "qty": "+126 GW/yr",
  "mode": "FORECAST",
  "role": "Cited-Firm",
  "context": "Morgan Stanley 'Powering AI' frames AI DCs as the single largest driver of global electricity demand acceleration.",
  "to_year": 2028,
  "conv_cues": "explicit annual rate; peer-to-sovereign comparison",
  "direction": "NUMERIC_TARGET",
  "from_year": 2026,
  "timeframe": "through 2028",
  "conv_level": "HIGH",
  "milestones": [
    {
      "kind": "llm_pre_event",
      "label": "Morgan Stanley publishes 'Powering AI' report with 126 GW/yr forecast",
      "notes": "HIT — Morgan Stanley insights article confirms 126 GW/yr AI DC growth equivalent to Canada's annual demand.",
      "source": "https://www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026",
      "status": "hit",
      "weight": 0.4,
      "ordinal": -14,
      "source_id": null,
      "confidence": 0.99,
      "source_url": "https://www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026",
      "expected_date": "2025-12-30",
      "observed_date": "2026-01-15",
      "research_origin": "deep_research",
      "expected_date_range": {
        "to": "2026-04-30",
        "from": "2025-09-01"
      },
      "measurement_criterion": "Morgan Stanley Research publishes 'Powering AI' (or equivalent) report quantifying AI DC power growth at >=100 GW/yr through 2028"
    },
    {
      "kind": "llm_pre_event",
      "label": "MS forecasts US DC demand reaches 74 GW by 2028",
      "source": "https://www.investing.com/news/stock-market-news/morgan-stanley-sees-up-to-20-shortage-of-us-power-for-data-centers-through-2028-4351251",
      "status": "hit",
      "weight": 0.4,
      "ordinal": -13,
      "source_id": null,
      "confidence": 0.97,
      "source_url": "https://www.investing.com/news/stock-market-news/morgan-stanley-sees-up-to-20-shortage-of-us-power-for-data-centers-through-2028-4351251",
      "expected_date": "2026-01-31",
      "observed_date": "2026-01-15",
      "research_origin": "deep_research",
      "measurement_criterion": "Morgan Stanley research publication forecasts US DC demand >=70 GW by 2028 with explicit power-shortfall warning"
    },
    {
      "kind": "llm_pre_event",
      "label": "Morgan Stanley Powering AI report reaffirms 126 GW/yr through 2028",
      "notes": "HIT — MS reaffirmed 126 GW/yr through 2028; framed as 1/5 of global power growth",
      "source": "https://www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026",
      "status": "hit",
      "weight": 0.4,
      "ordinal": -12,
      "source_id": null,
      "confidence": 0.95,
      "source_url": "https://www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026",
      "expected_date": "2026-02-28",
      "observed_date": "2026-02-28",
      "research_origin": "deep_research",
      "measurement_criterion": "Morgan Stanley publishes 'Powering AI' update reaffirming ~126 GW annual data-center power additions through 2028"
    },
    {
      "kind": "llm_pre_event",
      "label": "Morgan Stanley quantifies US shortfall — 49 GW by 2028",
      "source": "https://www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026",
      "status": "hit",
      "weight": 0.4,
      "ordinal": -11,
      "source_id": null,
      "confidence": 0.95,
      "source_url": "https://www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026",
      "expected_date": "2026-02-28",
      "observed_date": "2026-02-28",
      "research_origin": "deep_research",
      "measurement_criterion": "Morgan Stanley Research forecasts US data-center demand 74 GW by 2028 with ~49 GW shortfall in available power access"
    },
    {
      "kind": "quartile_checkpoint",
      "label": "Q1 window check-in (25%)",
      "status": "pending",
      "weight": 0.05,
      "ordinal": -10,
      "source_id": null,
      "expected_date": "2026-07-23",
      "observed_date": null
    },

... (truncated)