As demand for AI explodes, so does the need for new infrastructure. But building a conventional data center can take three to five years – an eternity in the age of generative AI, where chip designs evolve every quarter and competitive advantage can hinge on milliseconds.

One Japanese firm, Getworks, thinks it has a faster answer. Rather than constructing massive hyperscale facilities from scratch, the company assembles modular, containerstyle data centers in factories and ships them out for near-instant deployment. It’s a plug-and-play model for the AI age – compact, mobile, and quick to install.

Getworks is not alone in recognizing the opportunity. Japan’s digital infrastructure is straining under a perfect storm of forces: surging AI workloads, growing cloud adoption, and aging corporate IT systems. As the government pushes for “digital transformation” and firms rush to modernize, the country faces a compute crunch. Modular data centers – faster to build, easier to cool, and more flexible in where they operate – offer one possible solution.

Speed, however, is only part of the appeal. Getworks also seeks to align with broader shifts in energy and infrastructure policy. Its container units are designed to run on distributed renewable sources, such as solar and biomass, and to be sited in regions with surplus power – rather than adding to the load in grid-congested cities. That decentralization model, if it scales, could support both carbon goals and resilience against grid strain or seismic disruptions.