Tokyo, May 13, 2026, 00:13 JST
Exeo Group and Internet Initiative Japan are teaming up to build and market edge data centers tailored for AI workloads. Exeo plans to roll out its “Exeo Edge Data Center Solution” beginning May 11. Unlike distant cloud hubs, an edge data center is a compact facility sited near where data is created or processed. Exeo
It’s all about timing. Firms want AI inference—feeding trained models data to churn out real-time results—close to where the action happens: on factory floors, near inspection gear, in research labs or at medical centers. That’s where lags, bandwidth bills and sensitive data headaches make cloud solutions clumsy.
Exeo and IIJ positioned their new service as an answer to growing calls for secure, low-latency data processing right at the company site. According to Internet Watch, IIJ’s DX edge Cool Cube targets on-premise AI applications—from research and development to manufacturing, quality checks, and both medical and healthcare settings.
Exeo, which Wealth Advisor calls a major player in telecommunications construction, is stepping closer to the center of Japan’s AI infrastructure push through the new alliance. Plans call for a demonstration and showroom setup at the Exeo Iwatsuki training center in Saitama, where both firms aim to jointly test AI infrastructure operations.
Exeo has rolled out a new menu with two layouts: modular and container-type. For the modular setup, IIJ’s DX edge Cool Cube brings together racks, power, and cooling. Exeo takes care of picking and installing the hardware, bundling everything into a single service.
IIJ’s Cool Cube? It’s modular—install a single rack or expand, linking several together. The system handles air cooling for 45 kilowatts per IT module, or switch to liquid cooling for up to 60 kilowatts. With liquid cooling, heat gets pulled out via coolant, not just fans—a key fix as GPU-dense servers, especially those running AI, push thermal limits.
The two firms are also eyeing a cheaper package, putting IIJ’s modular data center together with GPU servers that Exeo has refurbished. It’s a pragmatic move. As enterprise AI efforts graduate from pilot phase to real deployment, demand is rising for more robust power, cooling, and stricter data management.
Tokyo’s stock market didn’t open at the dateline. On May 12, Exeo finished 1.4% lower at 2,950 yen, pulling back from a 2024 peak of 3,058 yen earlier in the session. IIJ, by contrast, gained 1.6% to close at 2,950 yen, according to Yahoo Finance.
Competition is ramping up. Back in April, NTT Group outlined plans to boost domestic data center IT power capacity from roughly 300 megawatts to 1 gigawatt by fiscal 2033. Over at NTT Data Group, an approximately 250 MW campus is on the table for the Inzai-Shiroi area in Chiba. Exeo and IIJ, meanwhile, are taking a different route, targeting smaller and more distributed deployments rather than focusing solely on sprawling campus sites.
This could be a deciding factor for factories, warehouses, or research centers where waiting on a big facility, or sending sensitive data off-site, isn’t an option. According to IIJ, edge computing handles data right on local devices or nearby computing resources, slashing latency and trimming the volume that needs to travel to remote cloud systems.
Still, the plan carries risks. Adoption hinges on companies making the case for on-premise AI over cloud setups, lining up power and cooling for each location, and wringing value from used GPUs without running into reliability headaches. Exeo and IIJ have yet to reveal pricing, sales goals or the names of any early customers.
