Physical compute scaling will proceed at ~0.5 OOMs per year through 2027, driven by exponentially larger data-center construction.
Predictor: Leopold Aschenbrenner
Prediction text
Physical compute scaling will proceed at ~0.5 OOMs per year through 2027, driven by exponentially larger data-center construction. | Hyperscaler quarterly capex disclosures
Key catalyst: Hyperscaler quarterly capex disclosures
Watch events: Single-site >5GW power contracts; hyperscaler quarterly capex disclosures.
Resolution evidence
xAI Colossus 200K→1M GPUs; Stargate $500B over 4 years; Meta/MSFT capex each >$75B 2025 — cumulative physical compute 2x-3x annually.
Predictor: Leopold Aschenbrenner
Calibration plot (stated vs observed)
Evidence about this node from Leopold Aschenbrenner 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
This node isn't linked to a reference class. The Bayesian update applies without outside-view blending.
Probability over time
Milestone chain
- 2024-09-27overdueQ1 window check-in (25%)
- 2025-06-25overdueQ2 window check-in (50%)
- 2026-01-31hitHyperscaler 2026 capex consensus crosses $600B (above 0.5 OOM/yr trajectory)How: Top-5 hyperscaler aggregate 2026 capex guidance crosses $600B per analyst consensusSource: https://introl.com/blog/hyperscaler-capex-600b-2026-ai-infrastructure-debt-january-2026conf 97%Notes: HIT — January 2026 consensus reached $600B-$630B; 36% YoY growth above the implied 3x physical compute trajectory.
- 2026-03-22overdueQ3 window check-in (75%)
- 2026-04-25hitHyperscaler 2026 capex revised upward to $720-750BHow: Q1 2026 earnings cycle prompts top-5 hyperscaler capex revision to $720B+Source: https://finance.yahoo.com/markets/article/magnificent-7-earnings-rush-reveals-ai-spending-surge-with-hyperscaler-capex-set-to-reach-725-billion-in-2026-224901707.htmlconf 95%
- 2026-06-01 → 2026-12-31pendingFirst 1 GW single-cluster training facility operationalHow: Hyperscaler announces first 1 GW power-rated training cluster operational (Aschenbrenner 'GW per cluster by 2026' call)Source: https://pro.stockalarm.io/blog/situational-awareness-two-years-later — Aschenbrenner '1 GW per cluster by 2026' tracking as HITconf 85%
- 2026-12-31pendingAggregate FLOPS deployed crosses prior-year level by ~3xHow: Industry FLOPS-deployed estimate (Epoch AI / SemiAnalysis) shows ~3x YoY growth (=0.5 OOM)Source: Epoch AI compute trends + SemiAnalysis hyperscaler trackingconf 70%Notes: Aschenbrenner: 0.5 OOM/yr = 3.16x. Capex pace ($450B 2025 -> $725B 2026 = 1.6x) supports below-3x FLOPS growth, partial confirmation.
- 2027-01-01 → 2027-04-30pending2027 hyperscaler capex guidance >=$900B (extends 0.5 OOM/yr through 2027)How: Hyperscaler 2027 capex guidance issued in Q4 2026 / Q1 2027 earnings >=$900B aggregateSource: https://datacenterrichness.substack.com/p/hyperscalers-plan-630-billion-inconf 55%
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
Raw metadata
{
"trf": 0.41580700323259134,
"kappa": 0.6875,
"base_rate": null,
"predictor": "Leopold Aschenbrenner",
"total_llr": -1.2163953243244932,
"grace_days": 7,
"bayesian_v2": true,
"prior_logit": 0.5908989803561037,
"bayes_factor": "2.3:1 against",
"blend_reason": "no reference_class linked",
"inside_prior": 0.6435713869374784,
"kappa_source": "predictor_table",
"n_milestones": 3,
"blend_applied": false,
"contributions": [
{
"llr": -0.4054651081081644,
"kind": "quartile_checkpoint",
"kappa": 0.6875,
"label": "Q1 window check-in (25%)",
"weight": 0.05,
"strength": "weak",
"confidence": null,
"source_url": null,
"adjusted_llr": -0.278757261824363,
"expected_date": "2024-09-27",
"measurement_criterion": null
},
{
"llr": -0.4054651081081644,
"kind": "quartile_checkpoint",
"kappa": 0.6875,
"label": "Q2 window check-in (50%)",
"weight": 0.05,
"strength": "weak",
"confidence": null,
"source_url": null,
"adjusted_llr": -0.278757261824363,
"expected_date": "2025-06-25",
"measurement_criterion": null
},
{
"llr": -0.4054651081081644,
"kind": "quartile_checkpoint",
"kappa": 0.6875,
"label": "Q3 window check-in (75%)",
"weight": 0.05,
"strength": "weak",
"confidence": null,
"source_url": null,
"adjusted_llr": -0.278757261824363,
"expected_date": "2026-03-22",
"measurement_criterion": null
}
],
"evidence_kind": "metadata_milestone_miss_sweep",
"inside_source": "history_v2",
"inside_weight": 0.708935097737186,
"outside_weight": 0.291064902262814,
"posterior_prob": 0.4389627352111058,
"posterior_logit": -0.24537280511698534,
"predictor_brier": 0.04167,
"inside_posterior": 0.4389627352111058,
"blended_posterior": 0.4389627352111058,
"reference_class_id": null,
"total_adjusted_llr": -0.836271785473089,
"predictor_n_resolved": 3
}Network propagation neighbors
Top incoming (parents)
Edges that influence THIS node's belief
| Kind | Node | Their prob | P(c|s=T) | P(c|s=F) | Δ implied |
|---|---|---|---|---|---|
| killer | TK02 AI Compute Supply Shock (TSMC/Taiwan Disruption) | 12.0% | 0.050 | 0.700 | +0.125 |
Top outgoing (children)
Predictions THIS node influences
No outgoing edges.
Ticker exposure
Beneficiaries (21)
Prerequisites (5)
| Type | Pred | Title | Domain | Lag |
|---|---|---|---|---|
| correlate | S_COMPUTE_1GW_2027 | Compute: 1GW operational by Jun 2027 | compute_scale | — |
| correlate | S_HUMANOID_ENTERPRISE_2028 | Humanoid R2: 100K+ enterprise by Nov 2028 | humanoid_deployment | — |
| correlate | S_GRID_50GW_2027 | 50GW dedicated AI/data center grid by Dec 2027 | energy_grid_expansion | — |
| correlate | S_COMPUTE_100GW_2030 | Compute: 100GW national-scale by Dec 2030 | compute_scale | — |
| killer | TK02 | AI Compute Supply Shock (TSMC/Taiwan Disruption) | — | — |
Dependents (0)
| Type | Pred | Title | Domain | Lag |
|---|---|---|---|---|
| No dependents | ||||
Validations (1)
| Observed at | Status | By | Notes |
|---|---|---|---|
| 2026-04-29 | partial | thesis_timeline_v1.0_import | xAI Colossus 200K→1M GPUs; Stargate $500B over 4 years; Meta/MSFT capex each >$75B 2025 — cumulative physical compute 2x-3x annually. |
Raw metadata
{
"nia": false,
"qty": "~0.5 OOMs/yr",
"mode": "FORECAST",
"role": "Cited-Researcher",
"caveats": "Gated by power contracts and grid capacity.",
"context": "First of three OOM vectors: raw FLOPs/watt expansion plus GPU count growth, equating to ~3x physical compute per year.",
"to_year": 2027,
"conv_cues": "quantitative rate; historical extrapolation",
"direction": "NUMERIC_TARGET",
"from_year": 2024,
"timeframe": "2024-2027",
"conv_level": "HIGH",
"milestones": [
{
"kind": "quartile_checkpoint",
"label": "Q1 window check-in (25%)",
"status": "overdue",
"weight": 0.05,
"ordinal": -6,
"source_id": null,
"expected_date": "2024-09-27",
"observed_date": null,
"miss_emitted_at": "2026-05-02T22:07:21.384228+00:00",
"miss_emitted_by": "metadata_milestone_sweep"
},
{
"kind": "quartile_checkpoint",
"label": "Q2 window check-in (50%)",
"status": "overdue",
"weight": 0.05,
"ordinal": -5,
"source_id": null,
"expected_date": "2025-06-25",
"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": "Hyperscaler 2026 capex consensus crosses $600B (above 0.5 OOM/yr trajectory)",
"notes": "HIT — January 2026 consensus reached $600B-$630B; 36% YoY growth above the implied 3x physical compute trajectory.",
"source": "https://introl.com/blog/hyperscaler-capex-600b-2026-ai-infrastructure-debt-january-2026",
"status": "hit",
"weight": 0.4,
"ordinal": -4,
"source_id": null,
"confidence": 0.97,
"source_url": "https://introl.com/blog/hyperscaler-capex-600b-2026-ai-infrastructure-debt-january-2026",
"expected_date": "2026-01-31",
"observed_date": "2026-01-31",
"research_origin": "deep_research",
"measurement_criterion": "Top-5 hyperscaler aggregate 2026 capex guidance crosses $600B per analyst consensus"
},
{
"kind": "quartile_checkpoint",
"label": "Q3 window check-in (75%)",
"status": "overdue",
"weight": 0.05,
"ordinal": -3,
"source_id": null,
"expected_date": "2026-03-22",
"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": "Hyperscaler 2026 capex revised upward to $720-750B",
"source": "https://finance.yahoo.com/markets/article/magnificent-7-earnings-rush-reveals-ai-spending-surge-with-hyperscaler-capex-set-to-reach-725-billion-in-2026-224901707.html",
"status": "hit",
"weight": 0.4,
"ordinal": -2,
"source_id": null,
"confidence": 0.95,
"source_url": "https://finance.yahoo.com/markets/article/magnificent-7-earnings-rush-reveals-ai-spending-surge-with-hyperscaler-capex-set-to-reach-725-billion-in-2026-224901707.html",
"expected_date": "2026-04-30",
"observed_date": "2026-04-25",
"research_origin": "deep_research",
"measurement_criterion": "Q1 2026 earnings cycle prompts top-5 hyperscaler capex revision to $720B+"
},
{
"kind": "llm_pre_event",
"label": "First 1 GW single-cluster training facility operational",
"source": "https://pro.stockalarm.io/blog/situational-awareness-two-years-later — Aschenbrenner '1 GW per cluster by 2026' tracking as HIT",
"status": "pending",
"weight": 0.4,
"ordinal": -1,
"source_id": null,
"confidence": 0.85,
"source_url": "https://pro.stockalarm.io/blog/situational-awareness-two-years-later",
"expected_date": "2026-09-15",
"research_origin": "deep_research",
"expected_date_range": {
"to": "2026-12-31",
"from": "2026-06-01"
},
"measurement_criterion": "Hyperscaler announces first 1 GW power-rated training clust
... (truncated)