AI Moves Up the Value Chain

A relay, not one trade

AI infrastructure and applications require different investment tests. Spending on chips and data centres is an input; durable customer revenue is the hoped-for output.

The International Energy Agency's scenarios show why electricity and infrastructure capacity are material constraints. But higher power demand does not establish the return on every data-centre project.

What software must prove

For an application provider, the relevant questions are whether AI raises retention, expands contract value or lowers the cost of serving customers. A feature used in a free trial is weaker evidence than a paid workload renewed at a healthy margin.

Alphabet reported that Google Cloud revenue rose 63% year over year to $20.0 billion in the first quarter of 2026, citing demand across AI and core cloud services. This is a company-reported result, not proof that every software supplier benefits equally.

Watch the handoff

Investors should track which part of the AI chain captures the economics as deployments mature. A supplier with scarce hardware can enjoy strong margins early. Over time, efficiency improvements or competition may shift value toward platforms with customers, data, distribution and workflow control.