Table of Contents
In its 2026 data center outlook, the ratings agency said hyperscaler-led AI expansion is driving unprecedented capital requirements
In sum – what to know:
$3 trillion investment outlook – Moody’s expects massive capital deployment over five years as hyperscalers expand AI-driven capacity.
Pre-leasing reduces oversupply risk – Most new capacity is already committed, though counterparty concentration risk is rising.
Costs and constraints intensify – Grid limits, local opposition, and rising construction and GPU costs are pushing data center expenses higher.
Moody’s Ratings expects global data center investment to exceed $3 trillion over the next five years, as operators race to expand capacity to meet surging demand for AI workloads.
In its 2026 data center outlook, the ratings agency said hyperscaler-led AI expansion is driving unprecedented capital requirements. Spending by six US hyperscalers—Microsoft, Amazon, Alphabet, Oracle, Meta, and CoreWeave—reached nearly $400 billion in 2025 and is projected to increase by a further $200 billion over the next two years.
To support this scale of growth, capital markets are evolving. Moody’s noted that the volume and diversity of development capital required have expanded significantly, with institutional investors increasingly lending alongside banks during the construction phase.
Despite the investment surge, Moody’s said the data center expansion remains in its early stages, with double-digit capacity growth expected to continue. Most new capacity is already pre-leased to hyperscalers, limiting the risk of widespread oversupply. However, the ratings agency warned that this trend is increasing counterparty concentration risk.
The report also highlighted a shift in how construction risks are allocated. Tenants are increasingly accepting delivery risks they have historically avoided to accelerate completion timelines, including exemptions related to power and utility availability. Moody’s said these changes are helping balance the risks associated with faster development.
At the same time, developers face mounting challenges. Local opposition to new data centers is rising due to concerns over power and water usage, while grid constraints are already restricting development in many regions. These pressures are partially offset by jurisdictions offering supportive regulatory and legal frameworks, particularly for AI-related projects.
Cost inflation remains another concern. Rising prices for construction equipment and GPUs are expected to persist through 2026, as increased production by miners and manufacturers remains insufficient to ease supply pressures. As a result, newer facilities are becoming more expensive, pushing lease rates higher.
Moody’s also cautioned that operational issues are likely to increase over time due to the rapid growth in data center numbers and the rising presence of newer, less experienced operators.
Similar projections have emerged elsewhere. A recent JLL report also forecast $3 trillion in investment, estimating that 100GW of new data center capacity will be added globally by 2030—effectively doubling existing capacity.
JLL estimated that AI-related uses could account for about half of total data center capacity by the end of the decade. Despite rapid growth, JLL says market fundamentals remain stable and do not point to a speculative bubble.
The JLL report describes this period as a major infrastructure investment cycle, shaped by power availability, grid delays, and rising construction costs.
