Liquid Energy

Companies that need AI inference capacity face a grid and construction bottleneck: demand for inference is growing faster than the power grid can keep up, and new datacentres often wait years for power hookups and permits before they can come online. Conventional datacentre builds are slow, one-off construction projects. This leaves companies of every size without a fast way to access new compute capacity, since traditional sites take years rather than weeks to deliver tokens.

Liquid Energy sells inference directly to companies, delivered through Nano Datacentres (NanoDC), self-contained units that ship complete from the factory. Each unit arrives, unfolds its own solar array, computes inference on site, stores energy, and comes online without any construction needed. Because each unit brings its own power, cooling, compute and networking, capacity can be added incrementally as a fleet rather than built once slowly, and new capacity can be placed wherever the sun is. The company states it can go from signed contract to first token delivered in under 8 weeks.

Demand for AI inference is growing faster than grid capacity, and new datacentres wait years for power and permits, creating an opening for power-delivering compute units that bypass the grid queue.