The four leading US hyperscale cloud computing giants have collectively raised their capital expenditure expectations for the latest fiscal quarter, signaling a rapid acceleration in AI infrastructure investment towards a trillion-dollar scale. According to analysis, Bank of America Securities reported on April 29th that Alphabet (Google), Microsoft, Amazon Web Services (AWS), and Meta Platforms, Inc. have all increased their capital expenditure outlook for fiscal year 2026 and beyond in their Q1 2026 earnings reports. Bank of America now forecasts that global hyperscale cloud provider capital expenditure will exceed $800 billion in 2026, a 67% year-over-year increase, and is expected to surpass $1 trillion in 2027, growing by approximately 25% further.
This wave of expenditure expansion is underpinned by the rapid monetization of AI services and sustained explosive customer demand. Each giant highlighted that accelerating AI revenue growth and the continuous realization of return on investment are the core drivers for increasing infrastructure investment. Concurrently, computing power supply is expected to remain tight throughout 2026, reinforcing the urgency for these companies to expand production capacity. For the semiconductor industry, this implies that AI chip suppliers, led by NVIDIA, along with sub-sectors such as memory, semiconductor equipment, power semiconductors, and optical components, will continue to benefit from this capital expenditure super-cycle.
The decision by these cloud giants to raise capital expenditure is not an indiscriminate expansion but is supported by robust business performance data. For Alphabet, its Gemini large language model now processes over 16 billion tokens per minute via direct API calls, a 60% sequential increase. Paid monthly active users for Gemini Enterprise grew 40% quarter-over-quarter, becoming a primary growth driver for Google Cloud in Q1. Google Search revenue grew at 19%, its fastest pace in recent years, significantly fueled by AI-driven search queries.
Microsoft reported that its AI-related sales have reached an annualized run rate exceeding $37 billion, a 123% year-over-year increase. Management anticipates that Azure cloud services will experience moderate acceleration in the second half of 2026, despite ongoing supply constraints. Amazon.com's AWS segment saw Q1 sales grow 28% year-over-year to $37.6 billion, its fastest divisional growth rate in over three years. The company attributes this demand to both its own workloads and partnerships, such as the Bedrock collaboration with entities like OpenAI and Anthropic.
Looking at specific figures, all four giants have significantly raised their projected 2026 capital expenditure compared to prior market expectations. Alphabet's Q1 capital expenditure was $35.7 billion, meeting expectations, but its full-year guidance was raised to approximately $185 billion, up from a previous $175 billion, with a预告 of a "significant increase" in 2027. Microsoft's Q1 capital expenditure was slightly below expectations, influenced by quarterly timing, but its full-year guidance was raised to $190 billion, a 61% year-over-year increase, far exceeding the prior market expectation of around $154 billion. Amazon.com's Q1 capital expenditure slightly exceeded expectations, and the company maintained its strong full-year guidance of over $200 billion, representing growth of more than 50%. Meta Platforms, Inc.'s Q1 capital expenditure was below expectations, also affected by timing, but it raised the midpoint of its full-year guidance to $135 billion, up from $1.25 billion.
Overall, Bank of America has raised its 2026 global hyperscale cloud capital expenditure forecast to over $804 billion, an approximately 7% increase from the previous expectation of about $750 billion, and anticipates further growth towards $1 trillion in 2027.
A significant point is that these companies have explicitly stated that increased raw material costs have been factored into their budgets. Microsoft specifically noted that its $190 billion 2026 capital expenditure expectation includes approximately $25 billion for component price increases, which accounts for about 70% of the increase over the prior market expectation ($154 billion). These cost increases pertain to key areas like memory, wafers, and substrates. Bank of America interprets this as a signal that AI semiconductor suppliers, in general, will maintain solid pricing power and profit margins, and that major computing and networking equipment vendors have the ability to pass these cost increases on to downstream customers.
The report also emphasizes that computing power supply will remain constrained throughout 2026, a key factor driving the continued capital expenditure expansion by hyperscalers. Regarding chip deployment strategy, the report notes that most hyperscale providers are equally prioritizing a heterogeneous mix of commercial GPUs and custom, in-house developed chips, rather than relying solely on one type of computing solution. This trend suggests that commercial GPU suppliers like NVIDIA and the proprietary chip projects of the cloud giants will develop in parallel, jointly addressing the massive computing demand. Bank of America identifies five main beneficiary categories within the AI semiconductor space: computing power (represented by NVIDIA), memory, semiconductor equipment, power semiconductors, and optical components.
Comments