Cascade Graph: Interactive Map of AI Energy Constraints and Supply Chokepoints

✍️ OpenClawRadar📅 Published: June 30, 2026🔗 Source
Cascade Graph: Interactive Map of AI Energy Constraints and Supply Chokepoints
Ad

The Cascade Graph is a directed knowledge graph of the physical economy, built to trace how structural pressure cascades from root drivers through chokepoints and geographies to investable tickers. It currently contains 405 nodes (34 drivers, 90 chokepoints, 10 geographies, 5 jurisdictions, 4 substitutes, 262 tickers) connected by 597 sourced mechanism edges and 41 feedback loops. Every link is in plain language, graded, and cited.

Stress flows left-to-right: drivers push pressure into the system, jurisdictions and geographies gate it, pressure concentrates at chokepoints, substitutes cap upside, and value accrues to investable terminals (tickers). Each ticker node carries 2–3 vetted instruments with sourced exposure notes, venue labels, and median daily $-volume liquidity tiers. Where no clean liquid play exists, the graph states that explicitly.

Four published theses trace full cascades:

  • The Uninsurable Future — climate losses → insurer retreat → catastrophe-bond & ILS market → RNR, AON, MMC, EG, ILS
  • The Copper Chokepoint — AI & grid demand → structural copper deficit → FCX, SCCO, COPX
  • The Nuclear Inevitability — AI power hunger → uranium & nuclear → CCJ, URA, URNM
  • The Northern Pivot — warming Arctic → new routes & resources

To use the graph, pick a theme from the dropdown to filter nodes, click any node to open its evidence file (why it matters, sourced data, tickers, further reading, and what it drives/drives it). Focus mode isolates a node's direct causes and consequences. Toggle between cascade view (one-way chains) and feedback loops (reinforcing or dampening cycles). The graph also tags each node as a problem play (own the scarcity), solution play (own what gets bought to fix it), or both.

Ad

The interactive version requires JavaScript. The site provides an accessible Cascade Index (plain text, no JS) for keyboard and screen reader users. All content is free, no sign-up, no paywall.

📖 Read the full source: HN AI Agents

Ad

👀 See Also

Single-call MCP pipeline reduces Claude Code token usage by 74%
Tools

Single-call MCP pipeline reduces Claude Code token usage by 74%

A developer built a context engine MCP server that provides Claude Code with a dependency graph of codebases, reducing token usage by 65% initially. A new single-call pipeline further cuts tokens by 74% by eliminating multiple round trips and deduplicating results server-side.

OpenClawRadar
Graphify: A Claude Code Skill That Built a Knowledge Graph of Your Repo — 450k Downloads, 40k Stars in 26 Days
Tools

Graphify: A Claude Code Skill That Built a Knowledge Graph of Your Repo — 450k Downloads, 40k Stars in 26 Days

Graphify is a Claude Code skill that reads every file in your repo, builds a knowledge graph with Leiden community detection, and queries it at 71x fewer tokens than raw files. 450k+ PyPI downloads, ~40k GitHub stars, #2 global rank in first week.

OpenClawRadar
Monarch v3: NES-Inspired KV Paging for 78% Faster LLM Inference
Tools

Monarch v3: NES-Inspired KV Paging for 78% Faster LLM Inference

Monarch v3 implements NES-inspired memory paging for transformers, achieving 78% faster inference (17.01 to 30.42 tok/sec) on a 1.1B parameter model with nearly zero VRAM overhead. The open-source algorithm splits KV cache into hot and cold regions with compression and promotion mechanisms.

OpenClawRadar
Building a Coding Agent for 8k Context: Planner/Executor Split, Token Budgeting, and Parallel Execution
Tools

Building a Coding Agent for 8k Context: Planner/Executor Split, Token Budgeting, and Parallel Execution

A detailed breakdown of building a CLI coding agent designed around 8k token limits, using a planner/executor architecture, strict token budgeting, and parallel task execution.

OpenClawRadar