Global investment in data centers reached a record US$ 61 billion in 2025, as per the findings of a recent S&P Global report. The surge reflects escalating demand for computing power driven by artificial intelligence, as cloud providers and infrastructure investors race to expand capacity.
The figure includes mergers, acquisitions, and direct spending on new builds and upgrades across major markets, according to S&P Global, marking the sector’s strongest year to date as reported by CNBC.
Hyperscalers such as Microsoft, Amazon, and Google remain central to this expansion, but the financing model is shifting. Rather than relying solely on internal cash, companies are increasingly turning to debt and external capital, tapping bond markets and private equity to fund large-scale, energy-intensive facilities. More than 100 transactions contributed to the total, underscoring how broad-based the investment push has become.
AI workloads drive construction boom
The scale of investment is connected to AI training and inference each of which require dense compute, advanced chips, and reliable power. Analysts describe the current cycle as a global construction rush with hyperscale campuses expanding in regions that can support heavy electricity use and offer regulatory clarity.
A McKinsey report forecasts that this is an early phase with global spending to scale data centers for AI hardware and physical infrastructure could approach US$ 7 trillion by 2030. Current investments, while record-setting, represent only a fraction of what may be required if AI adoption continues at its current pace.
Geography and power shape growth
In the United States, Virginia and Texas lead new development, benefiting from established infrastructure, tax incentives, and proximity to major networks. Internationally, Europe and parts of Asia are also attracting capital, driven by demand for low-latency AI services and local data requirements.
Power availability is a primary constraint which suggests AI-driven data centers could account for a substantial share of U.S. electricity demand by the mid-2030s, forcing developers to prioritize grid access, long-term power contracts, and on-site generation.
Ownership, financing, and energy pressures
Ownership models are evolving alongside financing with developers and operators taking on more of the build-and-run burden. While hyperscalers lease capacity or partner rather than owning every facility outright. This has broadened participation in AI infrastructure yet increased reliance on leverage.
Energy strategy has moved from a secondary concern to a core risk factor. Developers are investing in renewables, nuclear options, and advanced cooling to manage costs and regulatory pressure. Concurrently, concerns persist about grid strain, local opposition, and the possibility that AI returns may not fully justify the capital outlay nor result in significant return on investment (ROI).
Outlook: Growth with constraints
Utilization rates are rising, and deal activity shows little sign of slowing. Still, the sector faces limits tied to power, regulation, and financing risk. The US $61 billion milestone highlights both confidence in AI-driven growth and the structural challenges that will shape the next phase of data center expansion.
