GPU Cloud in Italy: NVIDIA RTX PRO 6000 and H200 in Milan (2026)

In short: OMC Cloud offers NVIDIA GPU servers in its Milan data center: the NVIDIA RTX PRO 6000 from $1.50 per hour and the NVIDIA H200 from $2.50 per hour, billed monthly. Data and models stay in Italy, latency from Italian cities is low and there are no egress fees. The NVIDIA B300 is also available on request, in limited quantities.

Why a GPU cloud in Italy

• Data in Italy: datasets, models and results stay physically in Milan, which matters for GDPR and for customers who ask where their data is.
• Low latency: in a measurement from 30 Italian locations, median latency to Milan was 12.8 ms, against 25.0 ms to Frankfurt. For chatbots, voice assistants and real-time inference the difference is noticeable.
• Predictable cost: a fixed monthly fee instead of hyperscaler on-demand rates, with no egress fees.

Which GPU to choose

• NVIDIA RTX PRO 6000 (96 GB GDDR7), from $1.50/hour: the entry point for inference of small and mid-size language models, generative AI, LoRA fine-tuning (for example Llama 3-8B) and real-time rendering.
• NVIDIA H200 (141 GB), from $2.50/hour: for the largest models, such as Llama 3-70B, high-memory training and long-context inference.
• NVIDIA B300 (288 GB): available on request in limited quantities, for frontier-scale models and serving several models on one machine.
• Other GPUs and multi-GPU setups: on request, with a quote from the sales team.

What is included

• Dedicated GPUs with full passthrough, not shared with other customers.
• CUDA, PyTorch and TensorFlow preinstalled; support for vLLM, TensorRT-LLM and Ollama.
• Gen 5 NVMe storage and no egress fees.
• The same 99.95% uptime SLA, ISO/IEC 27001, ISO/IEC 27017 and SOC 2 Type II certifications and 24/7 human support in English as every OMC Cloud server.

Frequently asked questions

Are the GPUs really in Milan?
Yes. The NVIDIA RTX PRO 6000 and H200 can be ordered in the Milan data center. OMC Cloud is part of the OMC Group, which owns Kamatera, and the Milan data center is one of the group’s infrastructure sites.

How is billing done?
GPU servers are billed monthly. Prices start at $1.50 per hour for the RTX PRO 6000 and $2.50 per hour for the H200.

Is the B300 available?
Yes, on request. Stock is limited: contact the sales team to check availability and get a quote.

Can I use several GPUs on one server?
Yes, multi-GPU setups are available on request with a dedicated quote.

Read more: GPU servers · VPS in Italy: cloud servers in Milan · Hourly VPS in Italy · Contact us