Liquid-cooled GPU colocation in West Texas
Behind-the-meter AI data center capacity purpose-built for NVIDIA B300 and GB300 NVL72 platforms — powered by on-site natural gas generation.
1 MW — 100+ MW capacity available — behind-the-meter GPU colocation in West Texas
Enquire →Advanced liquid cooling
Direct liquid cooling engineered for next-generation GPU platforms.
Behind-the-meter power
On-site natural gas generation with integrated power systems, built for AI rack densities.
High-speed connectivity
Networking sized for large-scale distributed training.
24/7 infrastructure operations
Continuous monitoring and on-site maintenance.
Speed to power, not a place in the queue.
Most AI colocation is gated by a utility interconnection that takes years to study, approve and build. CACTAVO does not wait in that queue. Power comes from on-site natural gas generation, behind the meter, with the fuel delivered on the property — so speed to power is measured in equipment lead times, not utility schedules.
The site sits in the Permian Basin of West Texas. Liquid-cooled AI data center capacity is available from 1 MW to 100+ MW, built for next-generation NVIDIA B300 and GB300 NVL72 clusters.
All electricity generated is consumed on site. The project is not designed to export power and does not interconnect with the transmission system.
Specifications
- Capacity available
- 1 MW — 100+ MW
- Module capacity
- 3.5 MW typical — 160 B300 Servers or 16 GB300 NVL72 Systems
- Platform
- NVIDIA B300 and GB300 NVL72
- Cooling
- Direct liquid cooling, 45°C high-temperature loop
- Power
- On-site natural gas generation, behind the meter — no grid interconnection
- Design standard
- Tier III design
- Construction
- Standardized modular units
- Operations
- 24/7 infrastructure monitoring and maintenance
- Location
- Permian Basin, West Texas, USA
Secure your AI capacity.
Tell us the capacity you need and when you need it, and we will come back with availability, rack density and commercial terms.