AI Infrastructure Economics
A filing-based business analytics project on how AI infrastructure demand appears across public-company value-chain positions.
This project asks one question: when AI infrastructure spending expands, which companies convert demand into profit, which companies carry capital burden, which companies remain cycle-exposed, and which companies mainly transmit AI demand?
Filing-based AI infrastructure research
Summary
Current answer
AI infrastructure is not one business model, and AI exposure alone is not enough to explain company economics. I read the current issuer-level comparison as suggesting that AI demand changes economic form as it moves through the value chain:
- Compute platforms can show operating leverage.
- Foundries and memory suppliers carry heavier capital burden.
- Memory remains exposed to pricing and inventory cycles.
- Equipment suppliers capture indirect bottleneck demand.
- Networking, power, and cooling vendors transmit AI buildout through different margin structures.
Issuer-level = whole-company filing metrics, not pure-AI revenue. The core analytical output is Profit Capture vs Capital Burden.
Navigate
Suggested reading path
01
Start with Profit Capture vs Capital Burden
The core analytical output — operating margin (margin capture) vs direct CapEx intensity (capital burden) across eight issuer-level companies.
Open comparison →02
Use company profiles for issuer detail
Each issuer profile documents filing-grounded metrics, role context, and interpretation boundaries for one company.
Jump to company profiles →03
Use Methodology & Data Policy for metric definitions and data boundaries
Metric definitions, issuer-level boundaries, validation rules, and data policy live on the methodology page.
Open methodology →
Outputs
Core outputs
Read the comparison first for conclusions and charts; use profiles and methodology for evidence detail and definitions.
Core comparison
Profit Capture vs Capital Burden
Eight-company scatter and evidence table for operating margin vs direct CapEx intensity.
Open core comparison →Company evidence
8 issuer-level evidence profiles
Filing-grounded issuer profiles for compute, foundry, memory, equipment, networking, and facility infrastructure.
View company profiles →Methodology & Data Policy
Metric definitions, issuer-level boundaries, validation rules
How metrics are defined, what issuer-level proxies mean, and where validation detail is documented.
Read methodology →Value chain
Value-chain role preview
Eight issuer-level company profiles in the current comparison set. Open a profile for filing-grounded detail; use the comparison page for the full economic map.
- NVIDIAmargin-capture compute platform
- AMDAI compute challenger
- TSMCcapital-heavy foundry bottleneck
- Micronmemory / cycle-risk profile
- ASMLlithography equipment bottleneck
- Broadcommixed custom silicon / networking / software profile
- Aristanetworking demand-transmission profile
- Vertivpower / cooling demand-transmission profile
Scope
This project is issuer-level, descriptive, and based on public filings. It does not provide valuation targets, trading recommendations, or pure AI revenue attribution. The demand-origin side of this spending — whether hyperscaler CapEx is translating into cloud revenue — is explored separately in the Cloud CapEx Translation module.