How does the AI economy operate?

The AI economy runs on three layers: the infrastructure that powers it, the models built on top of that infrastructure, and the applications and labor markets those models are reshaping, and understanding how money moves between those layers is what tells you where to position yourself.

The infrastructure layer: where the capital goes first

The United States is in the middle of its biggest capital-expenditure cycle in decades, dwarfing the telecom boom of the 1990s, and AI spending is spilling over into construction workers building data centers and the power industry installing turbines and transformers. The sector-wide spending by the major cloud companies, Meta, Google, Microsoft and Amazon, has reached roughly $600 billion in AI infrastructure capital expenditure.

That money flows into chips, cooling systems, real estate for server farms, and electricity. The companies that supply those inputs, semiconductor makers, utilities, construction firms, are the first to feel the boom, before a single consumer buys anything AI.

The model layer: where revenue is accumulating

External AI revenue is running at around $280 billion annualized. The top 10% of AI customers account for more than 90% of model and data-center spending, which means the money is concentrated, not spread evenly. A year ago, boards wanted to hear the AI strategy. Now they want AI unit economics. That shift, from whether AI works to whether it pays, is the through-line of how the industry is maturing.

AI products grew from 32% of revenue in 2025 to a projected 42% this year, and are on track for roughly 53% by 2027. The business model is mostly subscriptions and usage-based pricing: you pay per query, per token (think of a token as roughly three-quarters of a word processed by the model), or per seat per month.

The application layer: where most people will earn or lose

The center of gravity has moved toward the application layer. Close to two-thirds of builders are working on horizontal or vertical AI applications, with products targeting financial services and healthcare rising fastest, because a product embedded in a real workflow is harder to dislodge than a general-purpose model sitting one layer up.

This is the layer ordinary people interact with, and it's where the labor question gets sharp. In every modeled scenario, AI shifts income toward investors and away from workers, with the divide widening as the technology takes on more work. In the biggest-impact scenario, labor's share of income falls from 60% to 45% by 2030. The pain is concentrated among white-collar workers, with employment falling about 4% in a substantial scenario and more than 20% in an extreme one.

What this means for you

The people who thrive in this economy are not necessarily the ones who build the models. They're the ones who know how to use them to deliver outcomes in a specific domain, healthcare, law, logistics, creative work, finance. You don't need to build a trillion-dollar, all-purpose model to benefit from AI. But importing the tools and doing nothing with them is also not enough. What matters is adapting them to real local needs and building the skills to do more.

The clearest action: keep your skills current. CEEK Learning is where to pick up and refresh what you need to stay ahead of the curve.

If you're thinking about building something in this space, the Accelerators page has AI-specific funding, incubators and resources you can use right now.