Artificial Intelligence: Scaling the Intelligence Economy

AI investment is moving from model development toward a broad infrastructure and application cycle that could spread value across the economy.

• Global X estimates that AI infrastructure spending could reach $7.3 trillion between 2025 and 2030, with hyperscalers accounting for nearly 80%, while cumulative AI data center investment could reach $1.6 trillion over seven years.
• The supply chain extends from accelerators and servers to memory, networking, cooling, power equipment and data center construction; semiconductor sales could reach $1.5 trillion in 2026 and $2.3 trillion by 2030.
• Token demand could rise nearly 70-fold from 2025 to 2030 while global data center capacity grows 2.2 times, and available memory supply could be 18% below demand in 2030.
• Global X estimates that AI-related revenue from 15 companies could exceed the depreciation expense of AI assets by mid-2027 and overtake annual capital spending by late 2029, reaching about $1.6 trillion in 2030.
• A one-gigawatt AI facility requiring about $40 billion could generate roughly $12 billion in annual raw-compute revenue, $24 billion from inference, models and APIs, or $48 billion from end-user applications; AI software and applications could grow 7.8 times between 2025 and 2032.

Takeaway: The investment case is broadening from infrastructure providers to model, software and industry applications, with returns depending on whether growing capacity converts into measurable usage, revenue and productivity gains.