AI Capex Growth Continues

What happened

Hyperscalers, including Amazon, Microsoft, and Google, have significantly increased capital expenditure (CapEx) projections for AI and machine learning infrastructure. Collective investment is now forecast at $335 billion in 2025, a 16% rise from previous estimates. Amazon anticipates spending approximately $125 billion in 2025, with further increases in 2026. Microsoft's FY26 CapEx is projected to exceed its 45% FY25 growth, potentially doubling its data centre footprint over two years. Google raised its FY25 CapEx to $91-93 billion, signalling a substantial 2026 increase. This benefits semiconductor suppliers like NVIDIA and AMD.

Why it matters

The substantial increase in hyperscaler capital expenditure for AI infrastructure introduces a significant operational constraint for procurement and infrastructure planning teams. The rapid expansion of data centre footprints and compute capacity, driven by AI/ML demands, increases due diligence requirements for managing supply chain dependencies, particularly for specialised hardware from vendors like NVIDIA and AMD. This scale of investment also raises oversight burdens for IT Operations and Data Governance regarding resource allocation, energy consumption, and data security within rapidly evolving environments, increasing exposure to potential resource contention and integration complexities.