EN / 中文
Project Book

Invest

An evidence-gated radar for AI infrastructure bottlenecks and durable profit capture.

16 chapters

Contents

Open the book.

An evidence-gated radar for AI infrastructure bottlenecks and durable profit capture.

Begin with chapter 01

Chapter Index

Contents

01

Investment Constitution

Start from industry bottlenecks and independent structural reasoning; never treat hype, holdings, or a single opinion as the answer.

02

AI Architecture Map

Training, inference, server, rack, and data-center architectures determine the true constraint layers.

03

Bottleneck Discovery

Demand and architecture become investable bottlenecks only after passing supply rigidity, revenue conversion, and profit-capture tests.

04

Monopoly and Oligopoly Gate

Core candidates must explain the narrow market, customer lock-in, entry barriers, company position, and failure path.

05

Evidence System

Filings, earnings, and technical sources come first; social signals discover and challenge ideas, not replace facts.

06

Scoring and Classification

Score bottleneck importance, monopoly potential, consensus pricing, and discovery gap separately.

07

Compute and Host CPU

GPU racks still require host compute; CPU attach rates, procurement commitments, and architecture adoption determine monetization.

08

Memory, HBM, CXL, and DDR5

Memory demand is real, but crowding, capacity expansion, and control-layer profit capture must be separated.

09

Network and Photonics

Networking, interconnect, and photonics cross the observation gate only when architecture commitments, qualification, and product revenue appear.

10

Power, Cooling, and Infrastructure

Break power into transformers, switchgear, turbines, interconnection, busways, liquid cooling, and delivery timelines.

11

BOM, Materials, and Components

Generic component exposure is insufficient; AI-grade mix, lead time, supply concentration, and design-ins are the real levers.

12

Company Moat Dossiers

Company research must specify the business, customer lock-in, qualification, competitive structure, and bear case.

13

Core, Watch, and Warning Pools

Separate Core, Watch, Crowded Winners, Wait for Pullback, Low-Consensus Bottlenecks, and Narrative Avoidance pools.

14

Strategy Samples and Consensus

Treat Serenity, SemiAnalysis, and similar sources as strategy samples; every security still requires independent structural verification.

15

Weekly Radar History

Each radar run preserves date, evidence coverage, reasons for additions and removals, and rule changes.

16

Risks, Falsification, and Actions

Valuation, cycles, capacity, substitution, and execution failure must become an actionable falsification checklist.