Soluna Holdings, Inc. and Siemens have signed an MOU in order to take on one of the greatest bottlenecks of the AI revolution: the massive power demand swings of GPU-heavy workloads. As HPC scales, traditional grids cannot keep up with the rapid energy fluctuations that AI requires. This partnership focuses on the BTM environment, where data centers are powered directly by renewable energy sources.
The Texas Pilot: Project Grace
The core of the partnership is a 2 MW pilot project at Soluna’s “Project Grace” site in Texas. This site will serve as a testing ground for Siemens’ advanced electrical infrastructure, including:
SICAM SCADA Platform: For real-time monitoring and dynamic load response.
Advanced Hardware: Integration of Siemens transformers, switchgear, and power converters.
Load Validation: Testing how infrastructure handles the erratic power draw of modern AI chips compared to steady-state mining operations.
Also Read: NVIDIA Unveils Rubin AI Platform, Open Models & Autonomous Driving Innovations
Why This Matters for AI Infrastructure
“One of the biggest challenges to scaling AI technology is the amount of compute power it demands,” says John Belizaire, CEO of Soluna. The goal is to utilize “trapped” clean energy power that would otherwise be wasted to provide a low-cost, sustainable alternative for AI developers.
Brian Dula, President of Electrification and Automation at Siemens Smart Infrastructure USA, noted that the pilot is designed to create a “commercial-grade reference model.” By optimizing how renewable energy interacts with GPU demand, the two companies hope to provide a blueprint for carbon-neutral AI scaling.
The Strategic Takeaway
For enterprise leaders and IT architects, this partnership signals a shift toward decentralized, renewable-first AI infrastructure. As the global supply of GPU power remains constrained by energy availability, specialized BTM solutions like those being tested at Project Grace may become the standard for responsible, scalable AI deployment.
