FRAMEWORK · EVERGREEN
The 10-Layer AI Power Framework: A Complete Structural Guide
How compute, models, platforms, capital, regulation, and safety interact to determine who gains and loses power in AI.
→ Read analysis
L1 · COMPUTE & ENERGY
NVIDIA's Compute Monopoly: Why CUDA Still Owns AI
NVIDIA's FY26 revenue, the Blackwell sell-out, and CUDA's 90%+ share define the AI compute monopoly — and its five-year erosion timeline.
→ Read analysis
L4 · PLATFORM & INTERFACE
Platform Lock-In in AI: How Cloud Giants Capture Clients for Decades
AWS, Google Cloud, and Azure use compute agreements, proprietary APIs, and data gravity to build switching costs that outlast the model layer.
→ Read analysis
L7 · CAPITAL & MARKET
Capital Gravity: Why AI Money Always Flows to the Same Five Players
How frontier AI's compute economics created a self-reinforcing capital loop, and the three scenarios that could break it.
→ Read analysis
L1 · L8 · GEOPOLITICS
US-China Compute War: Bifurcation, Condominium, or Leakage
Export controls, Huawei Ascend, and sovereign compute programs are redrawing the global map of who can train frontier models.
→ Read analysis
L8 · REGULATION & GEOPOLITICS
Sovereign AI Programs: Country-by-Country Power Ranking
Ranked analysis of national AI programs by capital, compute, models, and regulatory architecture.
→ Read analysis
L2 · FOUNDATION MODELS
Data Network Effects: The AI Moat That Widens Every Day
Why behavioral data compounds into the most durable competitive moat in AI, and where the synthetic-data workaround breaks down.
→ Read analysis
L2 · FOUNDATION MODELS
Talent Concentration in the AI Labor Layer
Where elite AI research talent concentrates, and how that distribution shapes who can compete at the frontier.
→ Read analysis
L8 · REGULATION
Regulation Capture: How AI Incumbents Shape the Rules Governing Them
How the largest AI labs turned policy engagement into a competitive weapon — lobbying asymmetry, procurement moats, and the DoD split.
→ Read analysis
L3 · L4 · MIDDLEWARE
Enterprise Agent Control Plane: Who Owns the AI Autonomy Layer
Agent orchestration, governance standards, and the 18-month lock-in window forming around enterprise AI autonomy.
→ Read analysis
L3 · MIDDLEWARE
The MCP Protocol Power Shift: How Agent Middleware Is Reshaping AI
Anthropic's agent integration standard is redistributing power across the middleware stack — server growth, adoption, and who benefits.
→ Read analysis