AI’s capital expenditure shows no sign of cooling

Bank of America has significantly raised its estimates for global hyperscale spending on AI infrastructure, now projecting $611 billion by 2026, a $145 billion increase in just one month. Cloud giants including Google, Microsoft, Meta, and Amazon are aggressively expanding data center capacity, with capital expenditure intensity approaching 30% of sales, driven by the strategic imperative to scale compute power. This substantial investment is also fueling a 'second-order boom' in memory, chipmaking tools, and networking equipment, with BofA forecasting annual AI data center investment to triple to over $1.2 trillion by 2030, likening the scramble for capacity to an arms buildup.

Analysis

Bank of America has significantly upgraded its global hyperscale spending forecast for AI infrastructure, now projecting $611 billion by 2026, a substantial $145 billion increase from previous monthly estimates. This surge reflects aggressive capital expenditure by cloud giants, with Google allocating $92 billion in 2025, Meta targeting $100 billion in 2026, and Amazon planning to double its data center capacity by 2027. Capex intensity is nearing 30% of sales, approximately triple historic norms, indicating a strategic imperative for compute scale.

The extraordinary investment levels, approaching the strain seen during the 5G telecom buildout, are driven by the belief that scaling compute offers both offensive and defensive advantages in the AI race. This capital deployment is also catalyzing a "second-order boom" across the AI supply chain, benefiting memory, chipmaking tools, and networking equipment providers. BofA anticipates annual AI data center investment to triple to over $1.2 trillion by 2030, highlighting sustained demand.

The competitive landscape is characterized by an "arms buildup" for AI capacity, exemplified by OpenAI's $38 billion, seven-year deal with Amazon Web Services and strategic supply agreements with Nvidia, AMD, and Oracle. These commitments secure scarce chips and bandwidth, underscoring the critical importance of supply chain access. Analysts remain bullish on AI capex, expecting investments to nearly triple by 2030, with physical infrastructure and power capacity as potential constraints.

Key Decisions for Investors

• Investors should prioritize companies directly benefiting from the accelerating AI capital expenditure, including hyperscale cloud providers and critical hardware suppliers like chipmakers and memory producers.
• Monitor the sustainability of these high capex levels and potential supply chain constraints, particularly in chip availability, power infrastructure, and data center build-out capacity.
• Evaluate the long-term strategic positioning of companies within the AI ecosystem, focusing on those demonstrating clear competitive advantages through significant compute scaling and strategic partnerships.
• Consider opportunities in the "second-order boom" sectors, such as advanced memory, chipmaking equipment, and specialized networking solutions, which are experiencing robust demand driven by AI data center expansion.