Siemens & nVent unveil AI data centre cooling plan

Siemens and nVent have agreed a joint reference architecture for liquid cooling and power infrastructure aimed at large-scale NVIDIA-based artificial intelligence data centres.

The blueprint targets hyperscale facilities of up to 100 MW that run intensive AI workloads. It combines Siemens electrical and automation systems with nVent's liquid cooling technology and NVIDIA's DGX SuperPOD reference designs.

The companies are positioning the architecture for operators that are building or expanding AI clusters using systems such as NVIDIA DGX SuperPOD with DGX GB200. The design supports Tier III-capable configurations and uses a modular approach for rapid deployment across multiple sites.

Targeting hyperscalers

The reference architecture focuses on the rising power density of AI racks and the shift towards direct liquid cooling for graphics processing units and other accelerators. It sets out an integrated design for power distribution, automation, and liquid cooling infrastructure around those systems.

Siemens will supply industrial electrical systems and control technologies that support continuous operation of high-density compute halls. These include medium and low voltage distribution equipment and automation platforms that monitor and manage energy use.

nVent will contribute its liquid cooling solutions for server and rack-level heat management in high-density environments. The company has worked with major chipmakers and cloud providers on custom cooling designs for AI training and inference clusters.

As part of the collaboration, the firms intend to align their designs with NVIDIA's reference architectures for DGX SuperPOD installations. This creates a predefined layout and integration scheme for operators that adopt NVIDIA-based AI infrastructure at scale.

Focus on energy use

The partners are targeting improvements in both deployment speed and energy efficiency. They describe "tokens-per-watt" as a core metric for the architecture, which they define as AI output per unit of energy consumed.

"We have decades of expertise supporting customers' next-generation computing infrastructure needs," said Sara Zawoyski, President of nVent Systems Protection, nVent. "This collaboration with Siemens underscores that commitment. The joint reference architecture will help data center managers deploy our cutting-edge cooling infrastructure to support the AI buildout."

The design emphasises modular and fault-tolerant configurations that can be repeated across multiple data halls and locations. This approach aligns with the roll-out strategies of hyperscale cloud and AI providers that build standardised facilities in multiple regions.

Operators face growing pressure to reduce energy use and emissions while supporting rapidly expanding AI clusters. Many facilities now run racks that exceed traditional air-cooling limits, which has pushed a shift towards liquid cooling systems that sit closer to the heat source.

Reference architectures provide a template for such environments. They define how power, cooling, monitoring and control systems interface across vendors and technologies. They also create a consistent structure for supply chains and installation partners.

Siemens data centre push

Siemens has expanded its presence in the data centre market in recent years. The company applies automation, power distribution and building technologies from its industrial and infrastructure portfolios into mission-critical facilities.

It offers software-driven energy management and monitoring systems that connect with hardware in the power chain. These include sensors, protection devices, and digital services that track and optimise electrical performance in real time.

The firm positions its Intelligent Infrastructure business as a bridge between energy networks, buildings and industrial sites. Data centres sit in the middle of that strategy as both large energy users and key digital infrastructure assets.

Ciaran Flanagan, Global Head of Data Centre Solutions at Siemens, said the new blueprint targets faster AI deployments and energy-aware operation.

"This reference architecture accelerates time-to-compute and maximizes tokens-per-watt, which is the measure of AI output per unit of energy," said Flanagan. "It's a blueprint for scale: modular, fault-tolerant, and energy-efficient. Together with nVent and our broader ecosystem of partners, we're connecting the dots across the value chain to drive innovation, interoperability, and sustainability, helping operators build future-ready data centers that unlock AI's full potential."

nVent cooling strategy

nVent has built its liquid cooling business around direct engagement with chip manufacturers and original equipment makers. This includes cold plate and rack-level solutions for dense compute platforms as well as distribution units and manifolds that sit in the white space.

The company's portfolio spans cooling for servers, networking gear and power electronics. It also covers enclosure and protection products for industrial and infrastructure sites.

By aligning with Siemens and NVIDIA reference designs, nVent gains a defined role in upcoming AI data centre projects that follow the joint blueprint. The collaboration also creates a structured path for future product updates as chip thermal design power and rack densities rise.

The partners expect AI data centre designs to evolve further as workloads spread across more locations and as regulatory attention on energy use intensifies. They plan updates to the reference architecture as NVIDIA releases new AI systems and as operators refine their approaches to liquid cooling and power infrastructure at scale.