GPU Dedicated Server
Dedicated physical servers with exclusive resources, no sharing with others. From RTX-4090 to HGX H100, supporting full-scenario AI workloads from training to inference.
Dedicated Resources
No sharing with others, 100% hardware performance for exclusive use.
48-Hour Rapid Deployment
From order to launch, complete GPU cluster deployment within 48 hours.
Taiwan Data Sovereignty
Data centers located in Taiwan, all computing and storage within the territory.
Flexible Scaling
From single GPU to large clusters, supporting customized configuration.
GPU Dedicated Server Plans
Flagship GPU configurations, supporting DeepSeek-R1 full model deployment
RTX-4090 Professional
Best Seller- 8x NVIDIA RTX-4090 24GB
- AMD EPYC Processors x2
- 768GB DDR4 RAM
- 20MB Bandwidth (Unlimited)
RTX-5090 Flagship
Latest- 8x NVIDIA RTX-5090 32GB
- Intel Xeon Gold 6530 x2
- 512GB DDR5 RAM
- 20MB Bandwidth (Unlimited)
HGX A100 Enterprise
Enterprise Choice- 8x NVIDIA A100 80GB
- 1TB DDR4 RAM
- 200Gb InfiniBand x2
- 20MB Bandwidth (Unlimited)
HGX H100 AI Server
AI Training Flagship- 8x NVIDIA HGX H100 80GB
- 2TB DDR5 RAM
- 400Gb InfiniBand x8
- Supports DeepSeek-R1 671B
HGX H200 Super Server
Full Model Support- 8x NVIDIA H200 SXM5 141GB
- 2TB DDR5 RAM
- 400Gb InfiniBand x8
- Supports DeepSeek-R1 671B
GB200 NVL72 AI Super Server
New Flagship- NVIDIA GB200 NVL72 (72x Blackwell GPU)
- 13.5TB DDR5 RAM
- 800Gb InfiniBand x8
- Supports GPT-4o / Claude-3.5 and more
GB300 NVL72 Super Flagship
New Flagship- NVIDIA GB300 NVL72 (72x Blackwell GPU)
- 14.4TB DDR5 RAM
- 800Gb InfiniBand x8
- Supports GPT-4.5 / Gemini-2 and more
DeepSeek-R1 Model Selection Guide
Choose the right GPU configuration based on parameter size
| Parameters | Memory Required | Recommended Use |
|---|---|---|
| 1.5B | 1.1GB | Lightweight tasks, basic text processing |
| 7B | 4.7GB | Multi-domain applications, Q&A, dialogue generation |
| 14B | 9GB | Advanced language understanding, long-form generation |
| 32B | 20GB | Complex reasoning and writing |
| 70B | 43GB | Deep semantic understanding, creative writing |
| 671B | 404GB | Ultra-high precision reasoning, cross-domain tasks |
Memory formula: Memory ≈ Parameters (Billion) × 4 Bytes / Parameters
Three Steps to Launch
Requirements Interview
Confirm model scale, computing needs, project timeline and budget direction
Environment Setup
Assist with GPU selection, complete CUDA, Docker and required environment deployment
Final Delivery
Provide remote access information, online within 48 hours
Start Your AI Computing Journey
Our technical team can provide customized GPU solutions based on your needs