Bank of America: Global Cloud Giants' AI Capital Expenditure Could Surpass $1 Trillion by 2027

AI Arms Race Intensifies as the Trillion-Dollar Capex Era Arrives

After Google, Microsoft, Amazon, and Meta — the four U.S. tech giants — reported their Q1 2026 earnings and updated their capital expenditure outlooks, Wall Street has once again sharply raised its expectations for AI infrastructure spending. The semiconductor research team at Bank of America Securities released a new report raising its 2026 global hyperscale cloud capex forecast to over $800 billion, representing a staggering 67% year-over-year increase. Even more striking, Bank of America projects this figure will further surpass $1 trillion in 2027, growing approximately 25% year-over-year.

This means the global tech industry is pouring capital into AI infrastructure at an unprecedented pace and scale, and a new "trillion-dollar" era of AI investment is rapidly approaching.

All Four Giants Double Down, Capex Guidance Raised Across the Board

Analyst Vivek Arya's team at Bank of America noted that the key driver behind the forecast revision is the strong signals released by the four major hyperscalers in their latest earnings reports. Google, Microsoft, Amazon, and Meta all raised their full-year capex guidance during their respective earnings calls, demonstrating unwavering commitment to AI infrastructure buildout.

Based on their statements, demand for computing power for AI training and inference has far exceeded expectations, with investments in GPU clusters, custom AI chips, data center construction, and network infrastructure all accelerating. These tech giants broadly believe that the risk of "underinvesting" far outweighs the risk of "overinvesting," and that leadership in AI capabilities will directly determine the competitive landscape for years to come.

From Hundreds of Billions to Trillions: The Deep Logic Behind the Growth

The leap in global hyperscaler capex from the hundreds-of-billions level to the trillion-dollar level is driven by multiple factors:

First, AI model sizes continue to expand. From GPT-4 to Gemini Ultra and the next generation of even larger foundation models, the computing power required for training is growing exponentially. Training costs for each successive generation of models have surged dramatically, driving sustained demand for high-end GPUs and custom chips.

Second, AI inference demand is exploding. As AI applications move from experimental stages to large-scale commercial deployment, inference-side compute consumption is rapidly overtaking training. From intelligent search and AI coding assistants to enterprise AI solutions, every user interaction consumes substantial computing resources.

Third, data center construction has entered a "super cycle." To support massive AI workloads, tech giants are building and expanding data centers on a massive scale worldwide. Investments in supporting infrastructure such as power supply, cooling systems, and network architecture are rising in tandem.

Fourth, geopolitical competition is accelerating investment. The AI rivalry between the U.S. and China, along with AI strategic plans being rolled out by nations worldwide, is further fueling a "first-mover" mentality among tech companies at the infrastructure level.

The Semiconductor Supply Chain Enters a Super Boom

This Bank of America report was produced by its semiconductor research team, with a core focus on how trillion-dollar AI capex will reshape the semiconductor supply chain.

Undoubtedly, NVIDIA, as the undisputed leader in AI GPUs, stands to be the most direct beneficiary. Additionally, chip companies such as AMD, Broadcom, and Marvell, as well as foundry giant TSMC, will continue to enjoy order growth driven by the AI capex wave. Meanwhile, Google's TPU, Amazon's Trainium and Inferentia, and Microsoft's and Meta's custom chip initiatives indicate that in-house silicon will occupy an increasingly important position in the overall computing landscape.

At the data center infrastructure level, sub-sectors such as High Bandwidth Memory (HBM), advanced packaging, optical modules, and liquid cooling also face tremendous growth opportunities.

Market Concerns and a Rational Assessment

Despite the exciting growth outlook, cautious voices persist in the market. Some investors worry whether such massive capital expenditures can deliver commensurate returns on investment. The pace of AI commercialization, enterprise customers' willingness to pay, and macroeconomic uncertainties are all risk factors that require ongoing monitoring.

However, based on current statements from major tech giants, revenue growth in AI-related businesses is accelerating and the share of AI revenue contribution within cloud computing divisions continues to rise, providing reasonable justification for sustained high levels of investment.

Outlook: AI Infrastructure Investment Is Still in Its Early Stages

Bank of America's forecast reveals a clear trend: the global tech industry's investment in AI infrastructure is entering an entirely new order of magnitude. From $800 billion in 2026 to $1 trillion in 2027, while the growth rate decelerates from 67% to 25%, the absolute increase remains staggering.

More notably, if AI technology achieves more breakthrough advances over the next two years — for instance, substantial progress toward Artificial General Intelligence (AGI) or large-scale deployment of AI Agents in enterprise settings — actual capital expenditure could even exceed current projections.

Trillion-dollar AI capex is more than just a number. It signals that one of the largest technology infrastructure buildouts in human history is fully underway. For the entire tech supply chain, this represents both an unprecedented window of opportunity and a profound test of long-term value versus short-term exuberance.