# The NVIDIA H200 is a card made for data centres and designed specifically for AI/ML workloads, large language models, high-memory inference, and HPC.

*genai, semiconductor · news · 2025-12-22 · The Tribune*

## Key points

- CloudPe has integrated the NVIDIA H200 GPU into its platform for on-demand, cloud-native AI services.
- The H200 GPU features 141 GB HBM3e memory and 4.8 TB/s memory bandwidth, enabling much larger AI models.
- CloudPe offers highly competitive hourly pricing for H200 usage without minimum GPU bundle commitments.
- CloudPe enables Indian customers to rent H200 GPUs at ₹378/hour, avoiding hardware procurement and import costs.
- The platform targets startups, enterprises, and researchers who need flexible, scalable, memory-intensive AI compute.

**Companies:** NVIDIA, CloudPe
**Countries:** India

[Read the full story on The Tribune](https://www.tribuneindia.com/news/business/h200-gpu-launch-on-cloudpe-platform/)

---

Canonical: https://newsio.io/n/a8a50c47-b508-4a73-8ce1-1a137976d923/the-nvidia-h200-is-a-card-made-for-data-centres-and-designed-specifically-for-ai
Summarized by Newsio from The Tribune. https://newsio.io/how-it-works