Voip iconVoipSep 30, 2026 ~3 min source read

Submer launches Corenix: factory-built, Nvidia-based modular data centres for AI workloads

Corenix will factory-assemble high-density modules around Nvidia reference designs, then manage installation and commissioning to speed deployments for cloud, telco, and AI operators while confronting power, connectivity, and permitting limits.

Submer Unveils Corenix Modular AI Data Centres with Nvidia

Share this story

Send the public story page.

Useful takeaways from this story.

Corenix assembles fully tested, high-density data centre modules in factory conditions using Nvidia reference designs, then installs and commissions them on site to shorten delivery timelines.

Factory integration aims to improve repeatability and quality control, but grid capacity, local permits, fibre connectivity, and cooling resources remain potential bottlenecks.

What Corenix is and how it will work

Submer has launched Corenix, a new business that sells modular, AI-ready data centre platforms. Each Corenix product is a factory-built module that bundles IT, power, cooling, and networking into a single unit assembled and tested before shipment. The modules are built around Nvidia reference designs, which Submer says provide tested blueprints for high-density compute environments.

Why Submer chose factory-built modules

What Nvidia reference designs mean for buyers

Using Nvidia reference designs means Corenix will follow established layouts and component choices optimised for GPU-dense workloads. Those blueprints are intended to reduce planning uncertainty and accelerate validation of rack, interconnect, and cooling architectures for accelerated computing.

Modular delivery does not eliminate real-world constraints. Corenix still depends on local infrastructure: sufficient grid power, available fibre, water or other cooling supplies, and timely permitting. High-density sites place heavy demands on local electrical capacity, and those requirements often determine where and how fast new capacity can be brought online. Corenix's factory-tested modules may shorten on-site timeframes, but they cannot bypass regional infrastructure upgrades or regulatory approvals.

Leadership and customer-facing commitments

Martin Renkis will lead Corenix as CEO. His background includes senior roles at Johnson Controls and founding an IoT hardware and software firm, Smartvue. Corenix emphasizes a single-team responsibility model: the same group handles factory integration, testing, on-site commissioning, and operational readiness to reduce handoffs and improve accountability.

Implications for telcos and cloud communications

Faster, repeatable deployment of dense compute can change traffic patterns and edge demand. Telecom operators and cloud-communications providers may find it easier to plan capacity for low-latency services and AI-driven workloads if module delivery timelines are reliable. Still, operators will need to weigh power costs, site location, latency objectives, and total cost of ownership when deciding whether modular factory-built modules are the right fit.

Submer frames Corenix as a response to fast-growing demand for new compute capacity. The company stresses predictability and tested designs, but the practical effect will depend on execution and on resolving external constraints such as grid access and local permitting. Corenix aims to reduce lead times and deployment risk for customers who can meet the site-level infrastructure requirements.

More context around this story.

Loading more related stories...

Keep reading in the app

Open the app view to save this story, compare related coverage, and continue from the same source.

Open in app