What is gcube?
A GPU sharing cloud platform that connects developers who need GPU power with providers who have idle GPUs to spare.
Why gcube
High-performance GPUs are essential for AI model training, deep learning inference, and video rendering.
Yet public cloud costs remain prohibitive — while idle GPU capacity sits unused across the globe.
gcube solves both problems at once.
- Consumers — Cut GPU costs by up to 90% compared to traditional cloud providers
- Providers — Monetize idle GPUs and generate additional revenue
What you can run on gcube
| Workload | Description |
|---|---|
| AI/ML Training | Train and fine-tune deep learning models |
| LLM Inference | Serve open-source models with Ollama, vLLM, and more |
| Video Rendering | Run GPU-accelerated rendering pipelines |
| Jupyter Environment | Interactive notebook-based research and experimentation |
Core technology
Heterogeneous GPU Pooling
GPUs from different manufacturers and models (RTX 3090, H100, etc.) are unified into a single managed pool.
Choose the right GPU for each workload with full flexibility.
Container-Based Multi-Tenancy
Container isolation allows multiple users to share the same GPU simultaneously — securely.
Data security and environment consistency are guaranteed across all tenants.
Blockchain-Based Fair Compensation
A patented distributed contribution tracking system ensures transparent, fair payouts to every GPU provider.
Get started
- Run your first workload — Deploy a GPU workload in under 5 minutes
- Pricing & plans — Explore cost-efficient options across tiers
- Ollama + LLM serving — Serve open-source LLMs directly on gcube
Awards & recognition
- 🏆 CES 2026 Innovation Award — Recognized for global technology innovation
- 🔬 Selected for Korean Ministry of Science & ICT R&D Program
- 💰 Pre-Series A — ₩3B raised (Naver Cloud · BNK Venture Investment · TS Investment)