Compute (AI Capabilities)

📋Page Status

Page Type:AI Transition ModelStyle Guide →Structured factor/scenario/parameter page

Quality:0 (Stub)

Importance:0 (Peripheral)

Structure:

📊 0📈 0🔗 0📚 0•0%Score: 2/15

LLM Summary:This page contains only React component imports with no actual content about compute capabilities or their role in AI risk. It is a technical stub awaiting data population.

Critical Insights (4):

Quant.ASML produces only ~50 EUV lithography machines per year and is the sole supplier - a single equipment manufacturer is the physical bottleneck for all advanced AI compute.S:3.5I:4.0A:3.8
ClaimCompute governance is more tractable than algorithm governance: chips are physical, supply chains concentrated, monitoring feasible.S:2.2I:4.0A:4.0
Quant.TSMC concentration: >90% of advanced chips (<7nm) come from a single company in Taiwan, creating acute supply chain risk for AI development.S:2.5I:4.0A:3.0

Issues (1):

StructureNo tables or diagrams - consider adding visual content

Compute refers to the hardware resources required to train and run AI systems, including GPUs, TPUs, and specialized AI accelerators. The current generation of frontier AI models requires extraordinary amounts of computational power—training runs cost tens to hundreds of millions of dollars in compute alone. The significance of compute for AI governance stems from several unique properties: it is measurable (training runs can be quantified in FLOPs), concentrated (the global semiconductor supply chain depends on chokepoints like ASML, TSMC, and NVIDIA), and physical (unlike algorithms that can be copied infinitely, hardware must be manufactured and shipped).

Current Assessment

Level

35/100

Trend

↓declining

Confidence

70%

Compute concentration increasing; export controls having effect but circumvention growing

Last updated: 2026-01

Metric	Score	Interpretation
Changeability	30/100	Very difficult to influence
X-risk Impact	70/100	High direct x-risk impact
Trajectory Impact	80/100	High long-term effects
Uncertainty	35/100	Moderate uncertainty

What Drives Effective AI Compute?

Causal factors affecting frontier AI training compute. Note: This forms a cycle—AI capabilities drive revenue, which funds more compute—but feedback loops are omitted for clarity.

Expand

Computing layout...

React Flow

Node Types

Root Causes

Derived

Direct Factors

Target

Arrow Strength

Strong

Medium

Weak

Warning Indicators

Indicator	Status	Trend	Concern
Training run costs	$100M+ for frontier models	↓	medium
Chip concentration	TSMC produces 90%+ advanced chips	→	high
Export control effectiveness	Significant circumvention observed	↓	high

Addressed By

Intervention	Effect	Strength
Compute Governance	↑ positive	●●●
Export Controls	↑ positive	●●○

Scenarios Influenced

Scenario	Effect	Strength
AI Takeover	↑ Increases	strong
Human-Caused Catastrophe	↑ Increases	medium
Long-term Lock-in	↑ Increases	medium