Liquid cooling firm Submer has launched a new cloud unit.

The company this week announced the launch of InferX, a new unit it said will offer sovereign AI cloud services.

Submer said the new division aims to deliver a “faster, denser, and greener AI cloud built to accelerate the age of intelligence through a dual-plane strategy that connects core and Edge infrastructure.”

The company said core data centers will handle large-scale training and hosting in high-density GPU clusters, while Edge AI infrastructure will offer real-time inference close to users through telco and regional networks.

“AI is becoming a critical utility, and it deserves the same reliable, efficient, and sovereign infrastructure as energy,” said Daniel Pope, founder of InferX and Submer. "InferX has been created to provide that foundation – building an AI cloud platform that focuses on real AI use-case enablement."

The company said InferX – not to be confused with the US-based serverless inference platform of the same name – will combine Submer's liquid-cooling expertise with its advanced AI operations and delivery capabilities.

Details of any planned InferX facilities or deployments were not shared. On its website, InferX said it would use a vertically-integrated data center design featuring liquid-cooled GPU clusters.

Founded in 2015 by Daniel Pope and Pol Valls Soler, Submer has historically provided indoor and outdoor single-phase immersion cooling pods and larger-scale containerized pods.

Earlier this year, Submer announced plans to expand its remit and move into building data centers. The company’s first 56MW facility will be in Barcelona, Spain.

The company recently signed a Memorandum of Understanding with the government of Madhya Pradesh in India to jointly develop up to 1GW of liquid-cooled AI data centers – though it wasn’t clear if these would be operated by Submer or merely use the company as a cooling provider.