Hyperscalers are significantly increasing their capital expenditure (CapEx) plans to meet the growing demands of AI and machine learning workloads. Projections indicate a collective investment of $335 billion in 2025, marking a 16% increase from previous estimates. This surge is driven by the critical need to scale infrastructure, particularly data centres, to handle AI-driven services.
Key players like Amazon, Microsoft, and Google are leading this investment wave. Amazon anticipates spending approximately $125 billion in 2025, with further increases expected in 2026. Microsoft's CapEx for FY26 is projected to grow beyond the 45% increase seen in FY25, potentially doubling its data centre footprint over two years. Google has also raised its FY25 CapEx to $91-93 billion and signalled a substantial rise for 2026. This investment benefits semiconductor companies such as NVIDIA and AMD, who provide essential chips for AI computations.
Despite the current surge, some analysts predict a potential pullback in hyperscaler CapEx in 2026, suggesting a possible 20%-30% decrease based on historical investment patterns. This potential adjustment reflects a shift towards digesting the substantial AI infrastructure investments made in 2025. However, it is also suggested that compute demand will stay ahead of supply.
Related Articles

AI Capex: Risky Business?
Read more about AI Capex: Risky Business? →
Big Tech's AI Spending Spree
Read more about Big Tech's AI Spending Spree →
AI Investment: Market Appetite?
Read more about AI Investment: Market Appetite? →
AI Capex Boom Approaching Endgame
Read more about AI Capex Boom Approaching Endgame →
