Bank of America has identified similarities between the current artificial intelligence investment cycle and previous periods of large-scale capital spending, citing rising investment expectations, increased competition and declining prices for AI services.
In a note to clients on Friday, investment strategist Sebastian Raedler examined whether AI infrastructure spending could follow the pattern of earlier capital investment cycles, in which expanding capacity eventually placed pressure on profitability.
The bank highlighted the scale of planned investment by hyperscale technology companies, alongside recent developments in AI pricing and corporate spending.
AI Capital Expenditure Forecasts Reach $1 Trillion
According to Bank of America, consensus expectations for hyperscaler capital expenditure over the next 12 months have increased from $330 billion in early 2025 to approximately $1 trillion.
The latest forecast represents around 3% of US gross domestic product, exceeding the relative scale of investment recorded during two previous infrastructure and technology expansion periods.
The bank estimated that capital spending during the railroad expansion of the 1870s reached 2.2% of US GDP, while investment associated with the technology expansion of the 1990s represented approximately 1.1%.
Raedler said previous investment cycles frequently experienced declining financial returns as competition increased, even when the technologies involved continued to develop and gain adoption.
Competition and AI Pricing Developments
Bank of America identified increasing competition in the AI model market as a potential factor affecting the financial returns generated by infrastructure investment.
The bank noted that earlier capital investment cycles were often characterised by initially elevated profit margins, which encouraged additional investment and eventually contributed to excess capacity and reduced pricing power.
Raedler contrasted this pattern with the cloud computing market during the 2010s, when barriers to entry helped limit competition and supported industry profitability.
In the AI market, competition has increased at the model level, including from Chinese developers offering open-weight models.
According to the bank, average AI token prices have declined by approximately 50% since the summer.
The reduction in pricing has occurred alongside continued growth in the volume of tokens processed by AI systems.
Business AI Spending and Demand Trends
Bank of America also examined whether lower AI service prices could generate enough additional demand to offset the decline in revenue per unit of usage.
