Hyperscalers to Spend $1 Trillion on AI Infrastructure by 2027

JPMorgan CEO Jamie Dimon forecasts that hyperscaler spending will reach $1 trillion in 2027, more than doubling from $300 billion in 2025. This growth is largely driven by the increasing demand for AI infrastructure, with companies such as Amazon and Microsoft pouring billions into data centers, chips, power equipment, and networking gear.

The spending boom has already produced unprecedented demand across the AI infrastructure supply chain, driving orders for GPUs, high-bandwidth memory, networking chips, optical components, servers, power equipment, and cooling systems. This in turn is creating a powerful multiplier effect for semiconductor and networking companies.

Dimon also noted that some spending on AI investment may not have an easily measurable return, with improvements in customer experience and longer-term efficiency being harder to quantify. He added that heavy capital demand from infrastructure, remilitarization, and government deficits could push interest rates higher, while also warning of a potential market correction.