Each layer constrains and enables the layers above it. Power flows upward; constraints flow downward. Every daily signal is tagged to the layer where it originates.
L1
Compute Infrastructure & Energy
GPUs, chips, data centers, power grids. The physical foundation that constrains all AI above. Whoever controls compute controls the ceiling for every other layer.
What Controls Here
NVIDIA, AMD, and TSMC dominate chip supply. Nuclear and renewable energy determine data center capacity. Sovereign compute becomes a geopolitical tool.
Why It Matters
L1 decisions cascade upward. Energy shortages constrain model training (L2). Chip supply shapes which companies can build which models. Export controls fragment the layer into US/China/EU silos.
Key Players
NVIDIA, TSMC, Intel, AMD, Google (TPU), Meta (custom silicon), Huawei, energy grid operators, nuclear and renewable developers.
Signal Types to Watch
GPU availability shifts
Chip export controls
Data center power constraints
Sovereign computing initiatives
GPT, Claude, Gemini, and open-source alternatives. Whoever ships the base model shapes everything above. Closed vs. open, reasoning capability, and multimodal maturity define the competitive moats.
What Controls Here
OpenAI (GPT), Anthropic (Claude), Google (Gemini), Meta (Llama), and Chinese labs (Qwen, Baichuan) determine which models dominate training and inference workloads.
Why It Matters
L2 models are the gateway to every layer above. If a lab ships faster reasoning, every L3–L4 platform must adapt. If open-source models become production-ready, the competitive landscape fragments.
Key Players
OpenAI, Anthropic, Google DeepMind, Meta, xAI, Microsoft Research, Mistral, Hugging Face, Chinese labs.
Signal Types to Watch
Model capability breakthroughs
Closed vs. open shifts
Training cost improvements
Inference efficiency gains
Agent orchestration, protocols, embeddings, context windows. The hidden lock-in layer — switching costs rise sharply once a protocol is adopted at scale.
What Controls Here
Protocol and framework providers shape how developers integrate AI. Data pipeline layers determine the cost of training and inference.
Why It Matters
L3 looks technical but creates real market power. Once thousands of startups build on one protocol, switching to another becomes expensive — this controls distribution at every layer above.
Key Players
Anthropic, OpenAI, Hugging Face, LangChain, LlamaIndex, Ray, Databricks, Pinecone, Weaviate.
Signal Types to Watch
Protocol or standard adoption
Lock-in changes in pipeline layers
Agent framework consolidation
Embedding quality breakthroughs
ChatGPT, Claude.ai, Copilot, agentic OS. The consumer and enterprise gateway — where users meet AI. Platform control determines distribution and pricing power.
What Controls Here
The largest labs and device makers define which interfaces capture users and their data.
Why It Matters
L4 is where money flows directly from users. Whoever owns the relationship with 100M+ users owns the leverage — regardless of which model powers it underneath.
Key Players
OpenAI, Anthropic, Microsoft, Google, Apple, Meta, Perplexity, Together AI, Replicate, Hugging Face.
Signal Types to Watch
Platform user growth / retention
Pricing changes or tier launches
Agent OS announcements
Feature parity wars
L5
AI-native Applications
Consumer apps and enterprise SaaS built AI-first. The revenue layer, where AI creates ARR and proves return on investment.
What Controls Here
Startups and established SaaS players monetize AI at the user level across design, writing, coding, sales, and finance workflows.
Why It Matters
L5 proves that the layers below have real-world value. When an app ships AI features that keep users sticky, it validates the entire stack. Failed applications signal the promise underneath is not yet real.
Key Players
Midjourney, Runway, Cursor, Figma, Canva, Notion, NotebookLM, and thousands of indie startups.
Signal Types to Watch
Founder funding and exits
User growth rates
Feature adoption metrics
Churn and retention shifts
Physical AI, robotics, enterprise automation, industry-specific deployments. Where AI translates from digital to real-world ROI in manufacturing, healthcare, and finance.
What Controls Here
Robotics makers and enterprise integrators track how deeply AI penetrates manufacturing, healthcare, logistics, finance, and defense.
Why It Matters
L6 ROI determines whether the entire stack below survives. If AI can't prove physical productivity gains, capital dries up. Success here compounds back into capital inflows at L7.
Key Players
Boston Dynamics, Tesla, Figure, Sanctuary, Scale AI, ABB, KUKA, and enterprise software vendors like SAP, Oracle, and Salesforce.
Signal Types to Watch
Robotics deployment scale-up
Enterprise ROI announcements
Factory / facility automation wins
Workforce displacement signals
Funding rounds, M&A, IPOs, valuation signals. The acceleration layer — capital flows respond to signals below and fuel the next investment cycle.
What Controls Here
Venture capital, corporate venture arms, sovereign wealth funds, and hedge funds determine where capital flows next.
Why It Matters
L7 responds to the layers below but also drives investment back into them. When hundreds of billions flow into AI, chipmakers get the signal to expand capacity. When valuations crash, capex slows.
Key Players
a16z, Sequoia, Greylock, Bessemer, corporate venture arms of the largest tech companies, sovereign wealth funds, and public markets.
Signal Types to Watch
Mega-round funding announcements
M&A and acquisition patterns
Valuation multiple shifts
IPO timing and readiness
L8
Regulation & Geopolitics
The EU AI Act, US–China chip decoupling, national security reviews, bloc formation. The fragmentation layer — where politics redraw market boundaries.
What Controls Here
Governments and regulatory bodies determine which companies can operate where, who can export what, and how AI deployment is constrained.
Why It Matters
L8 fragments the global AI stack into geopolitical zones. A product that works in one bloc may be restricted in another. This layer shapes whether the market is one or many.
Key Players
US, EU, China, UK, India, and Japan governments, sovereign regulators, and trade organizations.
Signal Types to Watch
Export control announcements
Regulatory changes (AI Act, etc.)
Trade and tariff shifts
Bloc alignment changes
Alignment, environmental pressure, deepfake risk, systemic failure modes. The constraint layer — where safety and existential risk set hard limits on deployment.
What Controls Here
Alignment and safety researchers, and environmental engineers, determine acceptable risk thresholds.
Why It Matters
L9 does not accelerate growth — it limits it. When alignment risk becomes visible or environmental cost is too high, regulation follows fast. This is where every layer below hits its ultimate constraint.
Key Players
Leading AI labs' safety teams, independent alignment researchers, academic institutions, and environmental organizations.
Signal Types to Watch
Alignment research breakthroughs
Safety incidents or near-misses
Environmental impact reports
Deepfake or misuse incidents
Labor markets, education, wealth concentration, culture. The terminal effect layer, where AI reshapes human society — feedback loops flow back into regulation as political backlash.
What Controls Here
Labor economists, sociologists, policymakers, and educators track workforce displacement, inequality, education disruption, and cultural shifts caused by the layers above.
Why It Matters
L10 is not a power layer — it's an impact measure. Mass unemployment triggers political backlash at L8. Broadly shared benefits keep regulation permissive.
Key Players
Labor departments, education ministries, researchers, media, public discourse, and workers' organizations.
Signal Types to Watch
Workforce displacement reports
Wage and inequality data
Education system disruption
Cultural and media shifts