Global Data Center Demand Will Double by 2030
The shift toward agentic AI reasoning models is set to accelerate infrastructure needs as token consumption hits 118 quadrillion.
Updated on Sept. 23, 2026 in Data Centers
Live Poll
Will the surge in AI data center construction prove to be a sustainable long-term investment?
Global data center demand is projected to double by 2030, driven by an anticipated 24-fold increase in AI token usage. This growth reflects a significant shift as agentic and reasoning models begin to consume 15 times more tokens per request than standard generative AI.
Why it matters
The transition to agentic AI requires more intensive inferencing workloads that must be located closer to the point of task execution. This evolution is forcing a pivot toward standardized, modular data center designs to handle the projected scaling of compute infrastructure.
Inferencing workloads are projected to account for 60% of all data center activity by 2030. Over the last decade, AI performance and model complexity have grown 1 million times, while costs per token have dropped by a factor of 10 annually.
The players
Data Center Operators
Infrastructure entities managing the physical, electrical, and cooling stacks required to support large-scale AI model inferencing.
The details
Modern data centers are moving toward a five-layer architecture that integrates energy, specialized chips, physical infrastructure, models, and end-user applications. To improve efficiency and speed, operators are increasingly deploying standardized, modular building blocks. The systems also utilize advanced liquid cooling and DC power distribution to manage the thermal and electrical loads required by high-density compute clusters.
Timeline
2016 to 2026: Data center performance and complexity expanded one million times.
2024: Massive data center build-outs emerged across Europe and North Africa.
2030: Global data center demand is expected to reach double its current levels.
The Tech Race
The transition to modular, standardized data center builds represents a departure from monolithic, custom-designed facilities. This shift mirrors the rapid growth of agentic and reasoning AI model architecture, which now requires compute capacity that is physically closer to the user.
Users will see data centers increasingly move into their local regions as inferencing demand grows. This modular approach prioritizes speed and efficiency, which may reduce the latency of AI-driven applications and services.
The takeaway
The demand for compute capacity will be determined by the transition from generative AI to agentic reasoning models. Watch the build-out progress in Europe and North Africa as a primary indicator of whether modular infrastructure can keep pace with the projected 2030 demand.
Further reading
For more on the infrastructure supporting current compute trends, visit our Data Centers section.
Live Poll
Will the surge in AI data center construction prove to be a sustainable long-term investment?
