Morgan Stanley forecasts a significant cooling in AI-related capital spending, with growth dropping from 93% in 2026 to 14% by 2028. As the focus shifts from hardware to software integration, investors may track how corporate productivity gains and margin expansion evolve, alongside persistent infrastructure constraints like power shortages.
Morgan Stanley has projected a structural pivot in the global AI investment cycle, signaling that the initial phase of heavy hardware spending is set to moderate over the next few years. The research indicates that capital expenditure growth among hyperscalers—large cloud service providers—is expected to slow from an aggressive 93% in 2026 to approximately 14% by 2028. This transition suggests a shift in capital allocation from accumulating computing hardware, such as GPUs, toward the integration of AI-driven software and enterprise applications.
Hardware Growth Expected to Moderate
The anticipated cooling of capital spending marks a maturing phase for the AI industry. During the initial build-out, cloud providers and data center operators prioritized the purchase of expensive hardware infrastructure. As this capacity comes online, the focus is naturally transitioning to monetizing the investment through software applications. While demand for high-end semiconductors is expected to remain firm, the explosive growth rates seen in 2026 are likely to taper as corporations move from setting up the foundation to finding practical, revenue-generating uses for AI.
Power and Regulatory Constraints
Despite the expected slowdown in spending growth, the industry faces structural bottlenecks that could complicate the timeline. Power availability remains the most pressing hurdle, with reports suggesting a 57-gigawatt shortfall between 2026 and 2028. This limitation is a critical factor for data center expansion, as compute-heavy AI operations require consistent and massive energy supplies. Furthermore, regulatory hurdles have created headwinds for infrastructure growth. Estimates indicate that approximately USD 286 billion in US-based data center projects have been stalled or sidelined since 2025 due to local regulatory and environmental opposition. These constraints suggest that supply-demand imbalances for data center capacity may persist for several years, even as capital expenditure growth rates decelerate.
Impact on Enterprise Efficiency
As the cycle pivots toward software, the focus for investors is shifting toward verifiable return on investment and productivity. Companies that have successfully implemented AI technologies have reported an average 9.6% net increase in productivity over the past twelve months. Morgan Stanley anticipates that these efficiency gains will translate into roughly 100 basis points of net margin expansion by 2027. The sectors likely to capture the next wave of value include cybersecurity and infrastructure-related software, as these tools become core to business operations.
For Indian investors, this global shift has meaningful implications. Indian IT services firms have been at the forefront of helping global enterprises integrate new technologies. As the demand shifts from hardware procurement to software deployment and application development, these service providers may find new opportunities in consulting and integration projects. However, the outcome will depend on their ability to execute and provide measurable value to their clients. Investors may continue to track productivity metrics in the tech sector, software adoption rates, and updates on power capacity availability for data centers, which remains a key monitorable for the infrastructure industry.