---
1. Compute stack and governance fracture are co-equal structural hubs.
`AI Compute Stack Hegemony` and `Tripolar AI Governance Fracture` each carry 42 connections at weight 9 — the highest of any nodes. This is not coincidental: the compute stack is the material substrate of power, while the governance fracture is the political outcome. The graph encodes a claim that these two dimensions are symmetrically central, neither derivative of the other.
2. The US export control regime is the most undermined high-weight node in the graph.
`US AI Export Control Regime` has 29 connections at weight 8, but at least 13 distinct mechanisms point at it with the edge label `undermines` or `constrains`: `DeepSeek Efficiency Shock`, `DeepSeek Efficiency Paradox`, `DeepSeek Efficiency-Control Paradox`, `Huawei Ascend Independence Stack`, `China Military-Civil Fusion AI Pipeline`, `China Semiconductor Self-Sufficiency Drive`, `Open-Source AI as Geopolitical Weapon`, `AI Compute Governance Verification Gap`, `mBridge AI-Enabled Financial Warfare`, `AI Energy Cost Asymmetry`, `Rare Earth FDPR Mirror Strategy`, `EDA Software Chokepoint Dilemma`, `Global South AI Alignment Contest`, and `AI Energy Geopolitics Race`. No other high-weight node has a comparable ratio of undermining-to-enabling edges.
3. Three foundational nodes carry weight=1 despite very high connectivity.
`TSMC Geopolitical Chokepoint` (18 connections), `China-US AI Ecosystem Bifurcation` (20 connections), and `US-China Geopolitical Compulsion Mechanism` (18 connections) all carry weight=1. These appear to function as structural axioms — nodes that nearly every other mechanism references — rather than analyzed concepts. Their actual structural significance in the graph substantially exceeds their assigned weights.
4. The Global South is represented as a genuinely contested vector, not a resolved dependency.
`Global South AI Multi-Alignment` undermines `China-US AI Ecosystem Bifurcation` (w=7), while `China AI Surveillance Authoritarianism Export` undermines `Global South AI Multi-Alignment` (w=7), and `AI Safety Summit Diplomatic Architecture` also undermines it (w=6). No mechanism in the graph resolves this contest. The 2027-2035 lock-in window closes specifically for this node.
5. The EU's strategic position contains a structural contradiction encoded directly in the graph.
`Brussels Effect 2.0 AI Regulatory Power` is connected as a compensating strategy (`compensates_for` → `EU Digital Sovereignty Structural Trap`), while `EU AI Regulatory Self-Defeat Loop`, `EU Regulatory Trap Loop`, and `AI Regulatory Arbitrage Talent Vortex` simultaneously undermine the competitive position that makes regulatory export viable. The graph encodes both the strategy and its failure mode as active simultaneously.
---
Loop A: AGI Governance Vacuum ↔ First-Mover Race Logic (direct, 2-node)
- `AGI Governance Vacuum` --[amplifies, w=8]--> `AGI First-Mover Race Logic`
- `AGI First-Mover Race Logic` --[exploits, w=9]--> `AGI Governance Vacuum`
A direct reinforcing loop. Any increase in race logic reduces governance prospects; any reduction in governance increases race incentives. No balancing mechanism interrupts this loop in the graph.
Loop B: Taiwan Escalation (direct, 2-node)
- `Taiwan Silicon Shield Erosion` --[enables, w=9]--> `Taiwan Contingency AI Power Collapse`
- `Taiwan Contingency AI Power Collapse` --[amplifies, w=9]--> `Taiwan Silicon Shield Erosion`
Both edges are at weight=9. This is the highest-weight direct loop in the graph. The two nodes mutually reinforce each other with no balancing edges between them specifically.
Loop C: Export Control Self-Defeat (4-node)
- `US AI Export Control Regime` → (pressure on China's access) → `DeepSeek Efficiency Shock`
- `DeepSeek Efficiency Shock` --[triggers, w=9]--> `DeepSeek Efficiency-Control Paradox`
- `DeepSeek Efficiency-Control Paradox` --[amplifies, w=8]--> `Open-Source AI as Geopolitical Weapon`
- `Open-Source AI as Geopolitical Weapon` --[undermines, w=9]--> `US AI Export Control Regime`
The control mechanism produces the condition (efficiency pressure) that generates the instrument (open-weight models) that undermines the control. The loop also has a shorter path: `DeepSeek Efficiency-Control Paradox` --[undermines, w=9]--> `US AI Export Control Regime` directly.
Loop D: Safety Asymmetry Amplification (4-node)
- `AI Prisoner's Dilemma Structural Lock` --[amplifies, w=8]--> `China Safety Asymmetry in AI Race`
- `China Safety Asymmetry in AI Race` --[amplifies, w=8]--> `AI Governance Summit Entropy`
- `AI Governance Summit Entropy` --[amplifies, w=9]--> `AGI First-Mover Prisoner's Dilemma`
- `AGI First-Mover Prisoner's Dilemma` --[amplifies, w=8]--> `AI Compute Stack Hegemony`
- ...and separately: `AI Safety Multilateral Governance Collapse` --[amplifies, w=8]--> `China Safety Asymmetry in AI Race`
- `AGI First-Mover Prisoner's Dilemma` --[triggers, w=9]--> `AI Safety Multilateral Governance Collapse`
The race logic causes governance institutions to weaken, which increases safety asymmetry, which validates the race logic.
Loop E: EU Regulatory Self-Defeat (5-node)
- `Brussels Effect on AI Standards` --[triggers, w=8]--> `AI Regulatory Arbitrage Talent Vortex`
- `AI Regulatory Arbitrage Talent Vortex` --[drives, w=8]--> `EU AI Regulatory Self-Defeat Loop`
- `AI Regulatory Arbitrage Talent Vortex` --[amplifies, w=9]--> `EU AI Competitiveness Deficit`
- `EU AI Competitiveness Deficit` --[deepens, w=8]--> `Tripolar AI Governance Fracture`
- `Tripolar AI Governance Fracture` --[amplifies, w=9]--> `China-US AI Ecosystem Bifurcation`
- `China-US AI Ecosystem Bifurcation` --[amplifies, w=7]--> `EU Regulatory Sovereignty Trap`
The regulatory mechanism that is supposed to establish sovereignty generates the talent outflow and competitiveness gap that weakens the regulatory position.
Loop F: China Data Flywheel (3-node)
- `Inference Economy Supremacy Race` --[amplifies, w=9]--> `China Real-World Deployment Data Flywheel`
- `China Real-World Deployment Data Flywheel` --[amplifies, w=9]--> `Digital Silk Road AI Dependency Mechanism`
- `Digital Silk Road AI Dependency Mechanism` → deployment scale in partner countries → feeds back to `China Real-World Deployment Data Flywheel` (implicit through `China AI Surveillance Authoritarianism Export` --[amplifies, w=7]--> `China Real-World Deployment Data Flywheel`)
Wider deployment generates more real-world data, which improves models, which widens deployment.
---
1. EU sovereignty regulation enables Chinese soft power.
`EU AI Sovereignty Paradox` --[enables, w=7]--> `China Open-Source AI Soft Power Gambit`. The EU's restrictions on US AI models (applied through the AI Act) open market space for Chinese open-weight alternatives. The intended effect (blocking US dependency) produces the unintended effect (facilitating Chinese penetration).
2. India's third-pole strategy depends structurally on US export controls.
`India Third-Pole AI Strategy` --[depends_on, w=7]--> `US AI Export Control Regime` and `India AI Third Way` --[depends_on, w=7]--> `US AI Export Control Regime`. India's non-alignment positioning is only viable if US controls constrain China sufficiently to create space for differentiation. If controls erode (Loop C above), India's third-pole strategy loses a foundational condition.
3. AUKUS simultaneously enables and structurally excludes India.
`AUKUS Pillar II AI Defense Nexus` --[excludes, w=7]--> `India Third AI Power Emergence`, while `India Third AI Power Emergence` --[anchors, w=8]--> `Global South AI Multi-Alignment`. The primary US-led military AI alliance excludes the node that most anchors the non-aligned Global South dynamic. This is a structural incoherence between the military and soft-power dimensions of US strategy as represented in the graph.
4. China's demographic decline is encoded as a compulsion mechanism.
`China Demographic-AI Substitution Imperative` --[amplifies, w=8]--> `US-China Geopolitical Compulsion Mechanism`. Unlike strategic choices, demographic decline is not reversible on a decade-long horizon. The graph thereby encodes Chinese AI urgency as partially structural rather than purely elective.
5. Gulf sovereign wealth undermines the energy bottleneck.
`Gulf Sovereign Wealth AI Kingmaker` --[undermines, w=7]--> `AI Energy Bottleneck`. Gulf capital funding compute infrastructure in energy-abundant regions introduces a third supply-chain pathway that is neither TSMC-dependent nor Huawei Ascend-dependent, potentially providing an escape from both the US export control regime and China's digital silk road.
6. Safety asymmetry is an enabling condition for the AGI jackpot scenario.
`China Safety Asymmetry in AI Race` --[enables, w=8]--> `AGI First-Mover Geopolitical Jackpot`. In the graph's structure, willingness to develop AI without safety constraints increases the probability of first-mover advantage. This creates an inverse relationship between safety investment and the geopolitical jackpot outcome — slower, safer development reduces first-mover probability.
7. mBridge erodes export control enforcement infrastructure.
`mBridge AI-Enabled Financial Warfare` --[undermines, w=8]--> `US AI Export Control Regime`. Export controls rely in part on SWIFT-based financial system leverage for enforcement. The parallel settlement infrastructure directly reduces this enforcement lever, not just chip access.
---
`AI Compute Stack Hegemony` and `Tripolar AI Governance Fracture` (42 connections each, weight 9)
These two nodes function as the graph's primary attractors. `AI Compute Stack Hegemony` sits at the material layer — it is both an enabler (→ `AI Great Divergence`, → `AI Productivity-Power Conversion Mechanism`, → `Custom Silicon Race`) and a target of constraint from energy bottlenecks, export control erosion, rare earth leverage, and talent dynamics. `Tripolar AI Governance Fracture` sits at the political layer — it receives contributions from at least 18 different mechanisms and is rarely shown resolving or stabilizing. It functions as an absorbing state in the graph.
`US AI Export Control Regime` (29 connections, weight 8)
The graph's primary policy node. It is simultaneously the enforcement mechanism for US compute hegemony and the most-undermined node in the graph. Its high connectivity reflects its role as the hinge between US strategy and all the mechanisms that respond to it. The sheer number of undermining edges suggests the graph encodes a structural thesis: the export control regime is under multi-vector pressure that may exceed its capacity to adapt.
`China Open-Source AI Soft Power Gambit` (22 connections, weight 8)
Functions as China's primary soft-power instrument. It is enabled by `DeepSeek Efficiency Paradox`, `China Real-World Deployment Data Flywheel`, and `Authoritarian AI Structural Advantage`, and drives `Digital Silk Road AI Dependency Mechanism`, `AI Standards Multilateral Battleground`, and `Global South AI Infrastructure Alignment`. It is constrained by `Brussels Effect on AI Standards`, `India Digital Public Infrastructure Third Way`, and `AI Safety Summit Diplomatic Architecture` — all relatively lower-weight constraining edges compared to its enabling edges.
`Digital Silk Road AI Dependency Mechanism` (20 connections, weight 8)
The transmission mechanism by which Chinese AI capability translates into Global South alignment. It both triggers `Sovereign AI Movement` (as countries react) and amplifies `China-US AI Ecosystem Bifurcation` (as dependency deepens). Multiple mechanisms feed into it, and it serves as the bridge between Chinese technical capability and geopolitical leverage in the Global South.
---
1. Export control effectiveness vs. efficiency pressure.
The graph shows `US AI Export Control Regime` both as essential to US compute hegemony and as subject to 13+ undermining mechanisms. The net direction of these forces is not resolved. The graph records the pressure but does not model a threshold at which the regime becomes net-negative for US interests.
2. Brussels Effect applicability.
`EU AI Competitiveness Deficit` --[bets_on_replicating]--> `Brussels Effect on Textile Standards`. Simultaneously, `EU Regulatory Trap Loop` --[inversely_correlates]--> `Brussels Effect on Textile Standards` and `EU Strategic Autonomy Contradiction` --[undermines]--> `Brussels Effect on Textile Standards`. The graph records both the strategic bet and its potential failure conditions but does not specify what conditions determine which path dominates.
3. Taiwan deterrence: stabilizing or destabilizing?
`AI Drone Swarm Deterrence Paradox` both `amplifies` Taiwan Silicon Shield Erosion (destabilizing) and is described in its content as strengthening Taiwan's defensive deterrence. The same technology simultaneously raises the cost of attack and erodes the underlying silicon shield. The net deterrence effect is not resolved in the association structure.
4. China Semiconductor Self-Sufficiency as stabilizer.
`Taiwan Contingency AI Power Collapse` --[amplifies, w=8]--> `China Semiconductor Self-Sufficiency Drive` AND `China Semiconductor Self-Sufficiency Drive` --[undermines, w=9]--> `Taiwan Contingency AI Power Collapse`. If China achieves sufficient semiconductor independence, the strategic incentive for a Taiwan contingency partially diminishes (because the silicon shield becomes less valuable as a target). The graph records both directions but does not specify the threshold or timeline.
5. India's position contains unresolved structural dependencies.
India is simultaneously: anchoring `Global South AI Multi-Alignment`, excluded by `AUKUS Pillar II AI Defense Nexus`, dependent on `US AI Export Control Regime` for third-pole viability, and competing with `China Open-Source AI Soft Power Gambit`. These cannot all be simultaneously maximized. The graph does not specify which dependency binds first.
6. AGI Governance Vacuum has no closing mechanism.
Multiple nodes amplify `AGI Governance Vacuum`. No edge in the graph shows any mechanism closing or reducing it. The governance vacuum is represented as a destination but not a recoverable state.
7. `Chinese Government Veto Power` is a singleton with minimal integration.
`Chinese Government Veto Power` --[controls, w=7]--> `China Open-Source AI Soft Power Gambit`. This is its only edge. The node encodes a meaningful constraint on China's open-source strategy (state control over what gets released) but is not connected to the broader mechanisms that govern Chinese state decision-making.
---
H1: TSMC Geopolitical Chokepoint is the highest-consequence single-node failure point in the graph.
Despite weight=1, TSMC Geopolitical Chokepoint has direct dependency edges from `AI Compute Stack Hegemony`, `US AI Export Control Regime`, `Military AI Autonomy Race`, `Taiwan Contingency AI Power Collapse`, `US Techno-Alliance Architecture`, and `2027-2035 AI Power Lock-In Window`. A contingency event affecting this node would cascade through the largest number of high-weight nodes simultaneously. Testable prediction: interventions that reduce TSMC dependency (CHIPS Act, Intel expansion, Huawei Ascend scaling) should measurably reduce the weight of edges pointing into this node over time.
H2: The US export control regime will cross a self-undermining threshold before 2030.
Given Loop C (export controls → efficiency pressure → open-source models → undermines controls) and the 13+ additional undermining mechanisms, the regime faces multi-vector simultaneous pressure. If these mechanisms are partially additive, the enforcement overhead will at some point exceed the capability gap the controls are designed to maintain. Testable indicator: the gap between frontier US/EU models and Chinese open-weight models, measured on standard benchmarks, is the observable proxy.
H3: EU regulatory leverage is market-size contingent and exhibits a threshold effect.
The Brussels Effect works because EU market access is valuable enough that external firms comply with EU standards to enter. If `EU AI Competitiveness Deficit` deepens such that the EU becomes a net AI importer with limited local production, the threat of market exclusion loses credibility as an enforcement mechanism. Testable indicator: EU share of frontier model training and deployment relative to global share. Below some threshold (which the graph does not specify), the Brussels Effect mechanism fails.
H4: India's third-pole strategy has a 2027-2032 viability window.
`2027-2035 AI Power Lock-In Window` --[closes_for, w=7.5]--> `Global South AI Multi-Alignment`. India's strategy requires establishing independent compute, standards presence, and data infrastructure before the lock-in window closes. After that point, the graph implies that Global South countries will have made infrastructure alignment decisions that are difficult to reverse. Testable indicator: India's domestic GPU/TPU fabrication capacity and presence in ISO/ITU AI standards bodies by 2030.
H5: Loop A (AGI Governance Vacuum ↔ First-Mover Race Logic) is diverging, not converging.
Both edges in Loop A are high-weight (w=8, w=9), and no balancing mechanism interrupts the loop within the graph. If this structural reading is correct, international AI governance initiatives should show decreasing effectiveness over the 2025-2035 period, not increasing. Testable indicator: participation rates and binding commitment levels across the Bletchley → Seoul → Paris → subsequent summits.
H6: Gulf sovereign wealth introduces an unmodeled third compute supply chain.
The graph shows `Gulf Sovereign Wealth AI Kingmaker` funding both `Sovereign AI Movement` and `AI Compute Stack Hegemony` directly, while undermining `AI Energy Bottleneck`. If Gulf-funded AI infrastructure reaches scale independent of both TSMC and Huawei Ascend supply chains (using, for example, custom silicon fabbed in Japan or South Korea), this creates a compute pathway the current bipolar model does not account for. Testable indicator: Gulf SWF investment in non-US, non-Chinese semiconductor fabs and AI training clusters.
H7: China Safety Asymmetry creates a structural acceleration toward the AGI jackpot scenario regardless of US or EU policy choices.
`China Safety Asymmetry in AI Race` --[enables, w=8]--> `AGI First-Mover Geopolitical Jackpot`. Because safety asymmetry is itself enabled by `Authoritarian AI Structural Advantage` (a governance structure, not a technical parameter), the US and EU cannot close this asymmetry through technical means alone. The asymmetry is a political variable. Testable implication: arms-control-style verification regimes for AI development timelines face the same `AI Compute Governance Verification Gap` (w=7) that prevents nuclear-style treaties — compute is not physically observable in the way fissile material is.