Zutacore Logo White UpdateNeuralwatt-White (1)

 

Independently Validated AI Cooling Performance

Optimised Testing | ZutaCore & Neuralwatt

Download the results summary validating how two-phase, waterless direct-to-chip cooling eliminates thermal bottlenecks and unlocks measurable gains in AI performance, efficiency, and reliability.

<strong>EXPLORE THE VALIDATED RESULTS</strong>

 

 

 

This solution sheet breaks down Neuralwatt’s real test data to provide actionable insights for AI operators.

Initial findings include:

  • Zero thermal throttling under sustained AI workloads

  • 30% lower peak GPU temperatures

  • 22.5% improvement in tokens per watt

  • Zero thermal events during testing

  • Increased compute density without increasing power capacity

  • Reduced cooling energy overhead and operational risk

The results: verifiable performance improvements that directly impact throughput, efficiency, and system stability.

All achieved within waterless, high-temperature, production-ready AI facilities designed to scale with next-generation demand.

Testing conducted independently by NeuralWatt™ on Dell XE9640 cooled by ZutaCore® HyperCool® on 2x NVIDIA HGX H100 at Munters SyCool Data Center, Ireland. Additional trials underway.

ZutaCore & Neuralwatt Datasheet

About Neuralwatt™

Neuralwatt™ delivers real-time, energy-aware optimization at the compute layer - where all data center load originates. Neuralwatt's platform unlocks new revenue opportunities by maximizing performance per watt in the following ways:

  • Higher compute density – Run more inference per rack by intelligently tuning GPUs for token/$/watt gains.

  • Monetize stranded power – Flexibly scale workloads to use off-peak, curtailed, or demand-response capacity.

  • Thermal & sustainability gains – Enable compute to scale for cooling, heat reuse, and lower CO₂ per token.

  • Custom energy strategies – We integrate with unique environments.