Submer has chosen India as its manufacturing hub for Asia and set out plans to help shape a local ecosystem designed for AI-era computing. The company’s expansion, announced from Barcelona, frames India’s comparatively clean slate, fewer legacy cooling systems than developed markets, as an opening to build high-density facilities that are optimised for modern workloads from the start.
The move is explicitly tied to artificial intelligence. Submer positions its liquid cooling technologies as a route to support AI and other compute-intensive loads with lower energy and water use, and to reduce operational costs in facilities that must run dense accelerators reliably. “India’s digital transformation is unparalleled, and its lack of legacy infrastructure provides a perfect canvas for building the data centers of the future today,” Patrick Smets, CEO at Submer, said. The company’s aim, he added, is to “make India the model for how a nation can achieve massive data center scale while championing environmental stewardship.”
Liquid cooling aims to reset data centre design
The strategy emphasises liquid cooling as the enabling layer for AI-ready capacity. Submer states that its systems “drastically reduce energy and water consumption,” a claim that speaks to the pressure operators face as AI demand grows and grids tighten. By targeting chip-level heat extraction and high-density deployments, the approach is intended to decouple performance from the constraints of legacy air-cooled halls. In markets still early in their build-out, that shift can be designed in rather than retrofitted later at greater cost and complexity.
To accelerate adoption, Submer plans to work with original equipment manufacturers and original design manufacturers so that liquid cooling is integrated directly into high-performance compute infrastructure. That integration-first posture is meant to smooth deployment for operators who need to scale AI services quickly without bespoke engineering on each project.
The company also intends to establish a manufacturing facility in India. Beyond serving domestic demand, the site is envisaged as a production and export hub for its systems across Asia, positioning India as a supply point for regional AI data centre programmes. While timelines and locations are not specified, the manufacturing step signals a shift from market entry to local capability building.
Manufacturing and skills plan target regional scale
Submer’s plans extend to workforce development. The company says it will collaborate with skill development agencies and explore memoranda of understanding with state governments to roll out training at scale. The initiative is projected to create more than 5,000 mechanical, electrical and plumbing roles in the coming years, jobs that align with operating and maintaining high-density, liquid-cooled facilities. “By investing in skill development, we are not only ensuring the successful deployment of our technology but also creating long-term, high-value employment for Indian youth,” Dev Tyagi, President for the UK, Ireland and India, said.
For AI leaders and policy makers, the emphasis on skills is as consequential as the hardware. The operational discipline required for liquid systems, fluid management, leak detection, thermal monitoring and safe maintenance, demands new competencies on the data hall floor. Building that capability domestically is a prerequisite for sustained AI capacity rather than a sequence of pilot projects that struggle to scale. Submer’s stated focus on training signals recognition that technology choices and people choices must advance together if India is to build a resilient base for AI computing.
An ecosystem approach to AI-ready infrastructure
The company describes itself as a trusted advisor for sustainable, high-performance data centers, with solutions spanning cooling technologies and full-stack design and build, including AI-ready operations. Its India plan mirrors that positioning by combining technology integration with manufacturing and skills development. Submer notes it operates globally and, in 2025, expanded its solutions framework into four business pillars, Labs, Tech, Design & Build and AI Cloud, aimed at supporting the data centre lifecycle from R&D to advisory services.
For enterprises scaling AI, the direction of travel is clear in the details. Integrating liquid cooling with OEM and ODM partners shortens the path from procurement to production for dense inference and training clusters. Locating manufacturing in India supports quicker delivery cycles for regional builds. Investing in skills lowers operational risk in facilities where thermal performance is inseparable from service continuity. The combined effect is to make AI capacity less dependent on retrofits and more on designs that anticipate the heat and power profiles of modern models.
The announcement adds to a broader conversation about how new AI infrastructure should be built in markets still forming their data centre landscapes. By focusing on liquid cooling, domestic manufacturing and a trained workforce, the plan outlined here argues for an ecosystem model rather than piecemeal upgrades. If realised, it would give operators in India and across Asia a locally supported path to deploy AI workloads at scale while managing energy, water and cost pressures that are unlikely to ease.
Submer’s choice of India as a hub underscores a strategic bet: that designing for AI from the outset, rather than adapting facilities designed for earlier eras, will reduce friction as demand rises. For a sector where thermal limits increasingly define what is possible, the decision to couple technology with manufacturing and skills is as much about long-term resilience as it is about near-term capacity.
