Calcore builds AI‑native compute hubs in underutilized locations, offering high‑density racks (150 kW per rack) with ultra‑efficient designs (PUE 1.15) that integrate local energy sources, heat‑reuse and off‑grid power. These sites provide 20–50 MW of capacity (scalable beyond 100 MW) for fast, sustainable, and sovereign AI workloads, allowing customers to lease dedicated clusters with full hardware control and low latency.
Funding
Funding not disclosed
Founders
Product
Problem
AI developers and enterprises often face limited access to high-performance, energy-efficient compute infrastructure that offers both low latency and full control over hardware resources, especially in regions lacking traditional data center ecosystems.
Solution
Calcore constructs AI-native compute hubs in underutilized locations, delivering high-density, low-PUE racks that are optimized for the most demanding AI workloads. By integrating local energy sources, heat-reuse systems, and off‑grid power generation, the hubs provide fast, sustainable, and sovereign capacity. Each site is designed for rapid deployment, with initial power capacities of 20–50 MW and scalability beyond 100 MW depending on the location. Customers can lease ready‑to‑use AI clusters with a PUE of 1.15, ensuring maximum computational efficiency while maintaining full ownership of the hardware environment.
Target Audience
Primary customers are AI research labs, machine‑learning startups, and large enterprises that require dedicated, high‑performance compute clusters with full control over hardware and energy usage.
Features
- High‑density racks delivering 150 kW per rack for intensive AI training and inference
- Initial site power capacity of 20–50 MW with potential to exceed 100 MW per location
- Ultra‑efficient design achieving a PUE of 1.15 through heat‑reuse and off‑grid power integration
- AI‑optimized hardware configurations tailored for next‑generation models and workloads
- Rapid deployment timeline with the first operational site slated for Q2 2027
- Strategic placement in overlooked regions to provide low‑latency access and geographic sovereignty