Bank of America Predicts Global AI Capital Expenditure Could Surpass $1 Trillion by 2027

Bank of America Securities recently released a major semiconductor industry research report, significantly raising its capital expenditure forecasts for global hyperscale cloud computing companies and projecting AI-related capital spending to surpass $1 trillion by 2027. This astronomical investment scale marks the official entry of AI infrastructure construction into the "trillion-dollar era."

Big Four Tech Giants' Earnings Trigger Forecast Upgrades

The direct catalyst for this forecast adjustment was the successive release of Q1 2026 earnings reports by four major U.S. tech giants — Google, Microsoft, Amazon, and Meta — along with updated capital expenditure outlooks. After thoroughly analyzing the data, Bank of America Securities analyst Vivek Arya and his team made significant upward revisions.

Specifically, Bank of America raised its 2026 global hyperscale cloud computing capital expenditure forecast to over $800 billion, representing a year-over-year increase of 67%. Even more striking, the report projects this figure to climb further in 2027, breaking through the $1 trillion mark with approximately 25% year-over-year growth.

Looking at the growth trajectory, the 67% year-over-year growth rate in 2026 represents an explosive growth phase. While the 25% growth rate in 2027 shows some deceleration, the absolute increment remains staggering given the trillion-dollar base — meaning approximately $200 billion in additional investment within just one year.

Core Drivers Behind the Trillion-Dollar Investment

The sustained surge in capital expenditure is driven by demand converging from multiple dimensions.

Exponential growth in computing power demand. As large language model parameter scales continue to climb — from hundreds of billions to trillions and beyond — the computational resources required for training and inference are expanding at an unprecedented pace. Fierce competition among the Big Four tech giants in AI foundation models is directly driving massive procurement demand for GPU clusters, high-performance computing servers, and supporting network equipment.

Accelerating AI application commercialization. From enterprise AI assistants to code generation tools, from AI-powered search to intelligent customer service, AI applications are rapidly penetrating every industry. The acceleration of commercial deployment means continuously rising inference-side computing demand, compelling cloud computing giants to proactively build out data center capacity to meet customer needs.

Systematic upgrades to data center infrastructure. AI workloads impose far greater demands on power supply, cooling, and network bandwidth than traditional cloud computing. The large-scale deployment of new infrastructure such as liquid cooling systems, high-voltage power facilities, and ultra-high-speed interconnect networks is further driving up total capital expenditure.

Far-Reaching Impact on the Semiconductor Supply Chain

The trillion-dollar capital spending wave will have profound implications for the entire semiconductor supply chain.

First, AI chip manufacturers led by NVIDIA will continue to benefit. Demand for GPUs and custom AI accelerator chips (such as Google TPU and Amazon Trainium) will remain elevated, and advanced packaging capacity will continue to be tight. Second, segments including memory chips (especially HBM high-bandwidth memory), advanced-node wafer foundry services, optical modules, and networking equipment will also see significant order growth.

Moreover, this trend is reshaping the global semiconductor supply chain landscape. Foundry giants such as TSMC and Samsung are accelerating advanced-node capacity expansion, while competition among SK Hynix, Samsung, and Micron in the HBM space is intensifying.

Potential Risks and Market Concerns

Despite the seemingly bright outlook, the market is not without concerns.

Sustainability of return on investment remains the biggest question mark. Trillion-dollar capital investments ultimately need to be recouped through commercial revenue from AI products and services. If the monetization pace of AI applications fails to match the growth rate of capital expenditure, tech giants may face margin pressure.

Energy supply bottlenecks cannot be overlooked either. The power demands of large-scale AI data centers are placing enormous strain on electrical grids, with some regions already experiencing insufficient power supply — potentially becoming a critical bottleneck constraining the speed of data center expansion.

Geopolitical factors are also adding uncertainty. Ongoing issues such as chip export controls and supply chain security could affect the implementation pace and regional distribution of capital spending.

Outlook: The AI Infrastructure Race Is Far From Over

Bank of America's report reaffirms a core thesis: global tech giants' strategic bets on AI are not slowing down but rather continuing to intensify. From $800 billion in 2026 to $1 trillion in 2027, the scale of this AI infrastructure arms race has far exceeded prior market expectations.

For the entire tech industry, this signifies that AI is transitioning fully from an "exploration phase" into a "massive infrastructure buildout phase." Whoever can establish the most powerful computing infrastructure in this trillion-dollar investment race may secure a dominant position in the future AI ecosystem. The ultimate winners of this race will be determined not only by the scale of capital invested but, more importantly, by the ability to efficiently convert computing power into commercial value.