Runware introduces portable data center pod for AI inference

Runware, an AI infrastructure company, announced the launch of its modular data center called Sonic Inference Pod, a single transportable unit that represents a more flexible kind of compute. The Pod is designed to offer inference at higher quality and lower cost than other serverless inference platforms and GPU clouds, and its modular design allows for rapid capacity expansion without building fixed data centers. According to CEO Flaviu Radulescu, distributed compute positioned closer to end users will win in the long term, and the pods can scale quickly, deploy anywhere with power, and adapt to new hardware. The pods use a closed-loop cooling system that does not require water and can be built in days. Runware currently has 10 pods deployed across the US, Europe, and Asia-Pacific, serving companies like Higgsfield AI and Wix, with 160 sites available. The company raised $50 million in Series A in December. While AI labs like OpenAI and SpaceX are building massive data centers, Radulescu sees the pods as complementary, noting that if one pod goes offline, traffic moves to another, and customers who want dedicated hardware get entire pods. He also downplayed competition, citing the slow hardware design cycle and a small talent pool. Although building AI data centers is controversial due to resource use, Radulescu believes AI power use will increase regardless, and Runware focuses on meeting demand efficiently with no transmission losses, no water cooling, and using existing power capacity.