China Semiconductor and AI Industry Tracking

Problem

Tracking China’s semiconductor and AI ecosystem is noisy: old news recycling, thematic mapping, and vague “domestic substitution benefits” can easily污染 an investment thesis. This skill turns industry updates into verifiable, updatable research signals across equipment and core components, memory chips, AI compute infrastructure, and foundation models/AI applications.

How It Works

• Evidence ranking: prioritize company announcements, financial reports, filings, regulatory disclosures, tenders, and reliable industry data; treat media reports and rumors as supporting or unverified clues only.
• Materiality gate: include only updates that change demand, inventory, pricing, capex, orders, delivery, yield, software ecosystem, financing structure, or competitive position.
• Company mapping: distinguish direct mapping from verifiable indirect mapping, avoiding same-sector correlation, anonymous customers, stock-price moves, or sector rallies as order evidence.
• Cycle assessment: check short-term inventory cycles, mid-term innovation and capex cycles, and long-term domestic substitution trends.
• Hypothesis updates: use new evidence -> hypothesis change -> order/earnings impact -> valuation impact -> trading-position impact, rather than force-fitting every company into the report.

Boundaries

Useful for industry daily briefs, company tracking, order mapping, and hypothesis review; it does not replace primary due diligence or trading decisions. If there is no new material evidence, it should not produce a report for completeness. Improved industry sentiment also does not automatically imply improved company earnings.