AI Foundation / RESEARCH MAP 01
AI Software & Services
Who gets paid when an AI capability becomes a useful workflow?
From computing inputs to models and paid workflows.
Scope: Enterprise AI software and services; excludes a standalone valuation of chipmakers or cloud infrastructure.
Geography: Global business-model lens; US framework and vendor examples.
01 / HOW IT WORKS
Follow the value chain
DEX editorial map. Read left to right (top to bottom on mobile); companies may span several stages. Expand a stage to inspect its economics.
-
Compute & data
Cloud providers, data owners and tooling suppliers
Supply the resources used to build and run AI systems.
Revenue & bottleneck — Compute & data
- Revenue models to investigate
- Compute consumption, data licensing or tooling contracts.
- Bottleneck to test
- Can data rights and serving costs support the intended use?
-
Models & deployment
Model developers and deployment platforms
Turn resources into usable inference and evaluation services.
Revenue & bottleneck — Models & deployment
- Revenue models to investigate
- Usage-based APIs, capacity commitments or licenses.
- Bottleneck to test
- Does reliability hold under real workloads, not just benchmarks?
-
Applications & adoption
Software vendors, integrators and enterprise teams
Embed a capability into a customer's operating process.
Revenue & bottleneck — Applications & adoption
- Revenue models to investigate
- Subscriptions, implementation projects or usage fees.
- Bottleneck to test
- Will verified savings translate into repeat paid use?
02 / FOLLOW THE MONEY · DEX INTERPRETATION
The business-model lens
Follow customer revenue after inference, support and integration costs. A growing user count alone does not establish attractive unit economics.
03 / TEST THE THESIS · RESEARCH QUESTIONS
Signals to investigate. Risks to challenge.
Demand & adoption questions
- Which workflows have measurable time savings?
- Do pilots become recurring paid deployments?
What could weaken the thesis?
- Model errors, data rights and accountability may block adoption.
- Falling model prices may help customers without improving every supplier's margin.
04 / BUILD A WATCHLIST
Metrics worth tracking
Suggested research metrics, not measured values. Collect primary data with a date, geography and definition before making comparisons.
- Cost per completed task
- Include inference, retries and human review; compare like-for-like tasks.
- Paid retention
- Track continued paid use, not registrations or free trials.
- Task success rate
- Use a defined test set and disclose human intervention.
05 / CHECK THE EVIDENCE
Evidence anchors & sources
These sources support the statements below. They do not validate every editorial hypothesis, imply endorsement, or refresh the explorer's market estimates.
- Amazon Bedrock publishes input/output token pricing for several models, providing a concrete example of usage-based inference billing. [aws]
- NIST's AI RMF is voluntary guidance for managing risk across AI design, development, use and evaluation. [nist-ai]
- Company documentation
Amazon Bedrock Pricing
Supports: Billing-model example only; no price or market-size estimate is reproduced.
- Government framework
AI Risk Management Framework
Supports: Risk-management scope; not a revenue forecast or an official value-chain taxonomy.
CONTINUE THE RESEARCH
Go from map to evidence.
The 50-industry dataset is a separate screening resource with different coverage and source limitations. Related reports retain their own dates and scope.
Educational research framework only. Not investment, legal or medical advice.