Key Takeaways
- Salute now supports >20 GW of AI compute power across multiple continents, positioning it among the world’s largest NVIDIA‑partner infrastructure providers.
- The company’s Direct‑to‑Chip (DTC) liquid‑cooling service is the first end‑to‑end operational framework that blends proven procedures, dedicated training, and on‑site commissioning.
- DTC cooling delivers up to 30 % lower Power Usage Effectiveness (PUE) and 2× higher GPU density versus conventional air‑cooled racks.
- Salute’s NPN (NVIDIA Partner Network) status guarantees direct access to NVIDIA’s latest GPUs, software stacks, and co‑selling resources.
Salute Secures 20 GW of AI Infrastructure Capacity
Salute, an authorized member of the NVIDIA Partner Network (NPN), announced that it has booked contracts to deliver and manage more than 20 gigawatts (GW) of AI‑focused compute capacity worldwide. The contracts span data centers in North America, Europe, and Asia‑Pacific and are tied to some of the most demanding AI workloads—large‑language model training, high‑resolution video generation, and autonomous‑vehicle simulation.
Why 20 GW Matters
- Scale: 20 GW of AI compute equals roughly 8,000 NVIDIA H100 GPUs (each rated at 2.5 kW TDP).
- Revenue impact: Assuming an average service fee of $0.12 per kWh, the capacity could generate ≈$2.9 M per month in recurring revenue.
- Market positioning: Only a handful of global integrators exceed the 15 GW threshold; Salute now sits in the top‑5 tier.
Direct‑to‑Chip (DTC) Liquid‑Cooling Operations Service
Salute’s flagship offering is its DTC liquid‑cooling operations service, the first comprehensive package that couples:
| Feature | DTC Liquid Cooling (Salute) | Conventional Air Cooling |
|---|---|---|
| GPU density | Up to 2 × (e.g., 12 H100 GPUs per 42U rack) | 6–8 H100 GPUs per 42U rack |
| Power Usage Effectiveness (PUE) | 1.08–1.12 (30 % lower) | 1.30–1.45 |
| Thermal headroom | ±5 °C margin at 85 °C coolant inlet | ±2 °C margin at 40 °C air inlet |
| Installation time | 2 days per rack (pre‑tested loops) | 3–4 days (custom ducting) |
| Operational cost | ~$0.04/kWh (water‑based) | ~$0.07/kWh (fan power) |
| Reliability | MTBF > 250,000 h (sealed loops) | MTBF ≈ 120,000 h (fans) |
The service includes field‑tested operating procedures, certified technician training, and full‑scale commissioning to guarantee safe, 24/7 operation of high‑density, liquid‑cooled AI racks.
Customer Portfolio & Geographic Reach
Salute’s new contracts involve:
- Three hyperscale cloud providers in the United States, each deploying 4 GW of H100‑based clusters.
- Two European research institutions receiving 3 GW of AI compute for climate‑modeling workloads.
- One Asia‑Pacific autonomous‑driving consortium commissioning 5 GW of GPU farms for real‑time simulation.
All customers have elected Salute’s DTC service to meet strict thermal‑budget constraints while maximizing floor‑space efficiency.
Benefits to AI Providers
- Higher Throughput: By cooling GPUs directly at the silicon, the system can sustain full‑load power draw without throttling, cutting training time by up to 35 %.
- Lower Capital Expenditure: Higher rack density reduces the number of required cabinets by ~50 %, translating into $1.2 M savings in building and power infrastructure per 1 GW deployment.
- Sustainability: Reduced PUE and water‑recycling loops help customers meet EU Green Deal and U.S. ESG targets.
Salute’s NPN Advantage
Being part of the NVIDIA Partner Network grants Salute:
- Early access to NVIDIA DGX‑H100 and GH200 Grace‑CPU/GPU platforms.
- Co‑marketing and joint‑solution engineering with NVIDIA’s Global Solutions team.
- Direct technical support for CUDA, cuDNN, and AI‑framework optimizations.
These benefits accelerate time‑to‑market for customers and ensure that Salute’s cooling solutions stay aligned with NVIDIA’s hardware road‑map.
Bottom Line
Salute’s acquisition of contracts supporting over 20 GW of AI infrastructure underscores the rapid shift toward liquid‑cooled, high‑density GPU farms. By delivering a fully packaged DTC cooling service, Salute not only boosts compute density and energy efficiency but also provides a repeatable, low‑risk pathway for AI leaders to scale their workloads. With its NPN status and a growing global footprint, Salute is now a pivotal enabler for the next generation of AI breakthroughs.