US data-center demand will reach 74 GW by 2028 against a projected ~49 GW power-access shortfall — structural gap driven by retiring coal/gas, aging grid, and ~7-year interconnection queues.
Predictor: Morgan Stanley
Prediction text
US data-center demand will reach 74 GW by 2028 against a projected ~49 GW power-access shortfall — structural gap driven by retiring coal/gas, aging grid, and ~7-year interconnection queues. | 2026-27 interconnection approvals; firm-power contract announcements
Key catalyst: 2026-27 interconnection approvals; firm-power contract announcements
Watch events: PJM capacity auction results; nuclear restart timelines (Three Mile Island, Palisades); interconnection queue metrics
Resolution evidence
Confirmed via morganstanley.com 'Powering AI' (Feb 27 2026) and PJM capacity-market warnings; FT 'AI Power Crunch' visualization consistent.
Predictor: Morgan Stanley
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
Major grid expansion (>50GW) within 5y of announce
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
Milestone chain
- 2025-08-13hitMorgan Stanley Research formally publishes 49 GW power-shortfall forecast for 2028How: Morgan Stanley publishes research note quantifying about 20 percent 49 GW out of 74 GW US data-center power shortfall through 2028Source: Morgan Stanley - Powering AI: Markets Race to Invest in AI Energy Solutionsconf 99%Notes: HIT - direct anchor; Morgan Stanley Stephen Byrd publicly confirmed the forecast on YouTube and via investor notes.
- 2026-04-01hitHyperscalers commit 1T USD or larger cumulative AI infrastructure capex 2025-2026How: MSFT/META/GOOGL/AMZN cumulative reported AI infrastructure capex 2025+2026 1T USD or largerSource: Morgan Stanley capex tracking / Goldman Sachs 500B USD plus 2026 estimateconf 85%
- 2026-04-01 → 2027-12-31pending10 GW or more behind-the-meter natural gas turbine deals announced for data centersHow: Cumulative announced behind-the-meter natural gas / fuel cell deals reach 10 GW or more for AI data centersSource: Morgan Stanley energy / Bloom Energy fuel cell / nuclear data center deal trackerconf 75%Notes: MS sees natural gas turbines adding 15-20 GW supply fuel cells 5-8 GW nuclear 5-15 GW.
- 2026-06-01 → 2027-12-31pendingFERC / NERC publicly flags grid reliability risk in PJM/ERCOT/MISO from data center loadsHow: Regulator FERC NERC or RTO publishes formal reliability advisory citing data-center load as material grid risk in 2027-2028 windowSource: FERC / NERC reliability assessmentsconf 70%
- 2028-03-11pendingQ1 window check-in (25%)
- 2028-05-21pendingQ2 window check-in (50%)
- 2028-07-31pendingQ3 window check-in (75%)
- 2027-06-01 → 2029-12-31pendingFirst documented data-center AI capacity curtailment due to grid power shortfallHow: Top-10 hyperscaler publicly discloses delayed/curtailed AI training run or commissioning specifically due to grid power-access shortfallSource: Hyperscaler 10-K / 10-Q filings FERC interconnection queue reportsconf 55%Notes: Cascade - Morgan Stanley 49 GW shortfall implies actual curtailment events somewhere in 2027-28.
No downstream cascades — this prediction is a leaf in the dependency graph.
What if this resolves?
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
Network propagation neighbors
Top incoming (parents)
Edges that influence THIS node's belief
Top outgoing (children)
Predictions THIS node influences
No outgoing edges.
Ticker exposure
Beneficiaries (23)
Adverse (3)
Prerequisites (6)
| 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 | — |
| killer | TK01 | AGI Capability Plateau (2026-27 Training Stall) | — | — |
| killer | TK03 | AI Regulatory Moratorium (EU/US Capability Freeze) | — | — |
Dependents (0)
| Type | Pred | Title | Domain | Lag |
|---|---|---|---|---|
| No dependents | ||||
Expected milestones (1)
| Expected by | Description | Status |
|---|---|---|
| 2028-12-31 | [Energy/Regulation 2028-12] [INF_010] PJM capacity auction results; nuclear restart timelines (Three Mile Island, Palisades); interconnect [INF_012] IEA annual electricity outlook; FERC ISO load-growth filings; transformer supply chain r | pending |
Linked documents (10)
Raw metadata
{
"nia": false,
"qty": "74 GW demand / 49 GW shortfall",
"mode": "FORECAST",
"role": "Cited-Firm",
"context": "Morgan Stanley Research 'Powering AI' outlook (Feb 2026): single-site DC footprint scaling to 1-4 GW, natural gas, nuclear, off-grid solutions racing to close the deficit.",
"to_year": 2028,
"conv_cues": "sell-side specific GW quantification",
"direction": "NUMERIC_TARGET",
"from_year": 2028,
"timeframe": "by 2028",
"conv_level": "HIGH",
"milestones": [
{
"kind": "llm_pre_event",
"label": "Morgan Stanley Research formally publishes 49 GW power-shortfall forecast for 2028",
"notes": "HIT - direct anchor; Morgan Stanley Stephen Byrd publicly confirmed the forecast on YouTube and via investor notes.",
"source": "Morgan Stanley - Powering AI: Markets Race to Invest in AI Energy Solutions",
"status": "hit",
"weight": 0.4,
"ordinal": -10,
"source_id": null,
"confidence": 0.99,
"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": "2025-12-15",
"observed_date": "2025-08-13",
"research_origin": "deep_research",
"expected_date_range": {
"to": "2026-04-30",
"from": "2025-08-01"
},
"measurement_criterion": "Morgan Stanley publishes research note quantifying about 20 percent 49 GW out of 74 GW US data-center power shortfall through 2028"
},
{
"kind": "llm_pre_event",
"label": "Hyperscalers commit 1T USD or larger cumulative AI infrastructure capex 2025-2026",
"source": "Morgan Stanley capex tracking / Goldman Sachs 500B USD plus 2026 estimate",
"status": "hit",
"weight": 0.4,
"ordinal": -9,
"source_id": null,
"confidence": 0.85,
"source_url": "https://www.goldmansachs.com/insights/articles/why-ai-companies-may-invest-more-than-500-billion-in-2026",
"expected_date": "2026-06-16",
"observed_date": "2026-04-01",
"research_origin": "deep_research",
"expected_date_range": {
"to": "2026-12-31",
"from": "2025-12-01"
},
"measurement_criterion": "MSFT/META/GOOGL/AMZN cumulative reported AI infrastructure capex 2025+2026 1T USD or larger"
},
{
"kind": "llm_pre_event",
"label": "10 GW or more behind-the-meter natural gas turbine deals announced for data centers",
"notes": "MS sees natural gas turbines adding 15-20 GW supply fuel cells 5-8 GW nuclear 5-15 GW.",
"source": "Morgan Stanley energy / Bloom Energy fuel cell / nuclear data center deal tracker",
"status": "pending",
"weight": 0.4,
"ordinal": -8,
"source_id": null,
"confidence": 0.75,
"source_url": "https://www.morganstanley.com/insights/articles/powering-ai-energy-market-outlook-2026",
"expected_date": "2027-02-14",
"research_origin": "deep_research",
"expected_date_range": {
"to": "2027-12-31",
"from": "2026-04-01"
},
"measurement_criterion": "Cumulative announced behind-the-meter natural gas / fuel cell deals reach 10 GW or more for AI data centers"
},
{
"kind": "llm_pre_event",
"label": "FERC / NERC publicly flags grid reliability risk in PJM/ERCOT/MISO from data center loads",
"source": "FERC / NERC reliability assessments",
"status": "pending",
"weight": 0.4,
"ordinal": -7,
"source_id": null,
"confidence": 0.7,
"expected_date": "2027-03-17",
"research_origin": "training",
"expected_date_range": {
"to": "2027-12-31",
"from": "2026-06-01"
},
"measurement_criterion": "Regulator FERC NERC or RTO publishes formal reliability advisory citing data-center load as material grid risk in 2027-2028 window"
},
{
"kind": "scenario_signal",
"label": "Scenario fires: Compute: 1GW operational by Jun
... (truncated)