Global data center investment is entering a new phase. Capital spending on data center infrastructure could exceed $3 trillion by 2030, as AI drives demand for servers, accelerators, networking, storage, power and cooling.
- Global data center capex could exceed $3 trillion by 2030.
- AI infrastructure is the primary growth driver.
- GPU servers are only one part of the spending boom.
- Power, cooling, networking and storage are becoming major constraints.
AI Is Reshaping Data Center Spending
The rapid expansion of artificial intelligence is pushing data center investment toward levels that would have looked extraordinary only a few years ago.
According to a new Dell’Oro Group forecast, cumulative global data center capital expenditure is expected to surpass $3 trillion by 2030.
The expansion is being driven largely by the infrastructure required to train and run increasingly large AI models.
It Is Not Just About GPUs
AI accelerators receive most of the attention, but deploying thousands of high-performance GPUs requires an entire supporting infrastructure.
Operators need high-bandwidth networking, large storage systems, powerful host servers, advanced cooling and enormous amounts of electrical capacity.
Having access to GPUs matters, but companies also need enough power, cooling, networking and physical space to actually deploy them.
Power Is Becoming a Critical Constraint
Modern AI clusters can consume dramatically more electricity than traditional enterprise infrastructure.
As deployments move from individual GPU servers toward entire AI-focused campuses, access to electrical capacity is increasingly influencing where new facilities can be built and how quickly they can become operational.
This is also pushing operators toward liquid cooling and other high-density thermal-management technologies that were previously limited to specialized HPC environments.
Server Infrastructure Will Benefit Too
The investment boom extends well beyond specialized AI accelerators. Traditional server CPUs, memory, NVMe storage, Ethernet and high-performance networking equipment remain essential components of large AI environments.
For server vendors and data center operators, that means AI could continue driving demand across a much broader infrastructure stack through the end of the decade.
The projected $3 trillion data center investment wave shows how quickly AI is moving from a software story to a physical infrastructure story. The next phase of the AI race will depend not only on faster chips, but on who can build and power the data centers required to run them.
Source
The forecast is based on new data center infrastructure projections from Dell’Oro Group published in August 2026.
