SINES Data Campus

Liquid Cooling for AI

Built from the ground up for AI-native workloads, liquid cooling and high-density compute.

The SINES Data Campus integrates liquid cooling technologies into a flexible, scalable design powered by 100% renewable energy.

Scroll
At a glance

Density, proven in production

SIN01, the first building of the campus, already runs one of the European Union's largest NVIDIA deployments.

130 kW+

Rack densities supported by the NVIDIA GB300 NVL72 deployment at SIN01

12,600+

NVIDIA Blackwell Ultra GPUs deployed at SIN01 by Nscale for Microsoft

66,000+

NVIDIA Vera Rubin GPUs announced by Nscale, starting in late 2027

1.1

Design PUE target for the campus

AI deployments

Frontier AI at SIN01

In October 2025, Nscale selected Start Campus for one of the European Union's first deployments of the NVIDIA GB300 NVL72 platform, in support of Microsoft.

The deployment supports rack densities surpassing 130 kW, delivering the power and cooling required for advanced AI workloads.

In May 2026, Nscale announced an expansion with more than 66,000 NVIDIA Vera Rubin GPUs, starting in late 2027, building on over 12,600 NVIDIA Blackwell Ultra GPUs already deployed at SIN01 for Microsoft.

Digital twin of an AI data hall at the SINES Data Campus
Digital twin of Nscale's operations at SIN01, built with Schneider Electric, AVEVA and NVIDIA.
Cooling options

Cooling that scales with density

Liquid cooling is part of the campus design, not an add-on. SIN01's capacity was expanded in response to customer requirements through the use of liquid cooling technologies.

Liquid

Integrated liquid cooling

Liquid cooling technologies are integrated into the flexible and scalable design of every building on the campus.

Partner

JetCool microconvective cooling

Available to customers who need efficient, high-temperature liquid cooling for AI GPUs, using the campus's sea water cooling infrastructure and reducing the need for mechanical chillers.

Read the announcement →
Ocean

Sea water heat rejection

A once-through sea water circuit, reusing the infrastructure of a decommissioned power plant, is the design basis for rejecting heat at the campus.

Design before you build

A digital twin of the AI factory

In October 2026, Schneider Electric, AVEVA, Nscale and Start Campus piloted an end-to-end digital twin built with the NVIDIA Omniverse DSX Blueprint.

The twin simulates how a jump in rack density moves through the power train and the liquid cooling loop before it becomes a physical reality. Operator and customer can test new hardware generations, agree the configuration and build it once.

Certified for hyperscale

Built to hyperscale standards

CertificationScope
OCP Ready v2 HyperscaleSIN01 and SIN02, assessed by the Open Compute Project Foundation
LEED Gold BD+C v4SIN01, 75 points
ISO/IEC 27001SINES Data Campus, within the defined scope
SOC 2 Type II reportSINES Data Campus, within the defined scope
TIA Tier III designConcurrently maintainable infrastructure

Looking for AI-ready capacity in Europe?

Talk to the Start Campus team about capacity, timelines and delivery models at the SINES Data Campus.

Contact us →