Europe
Availability on request
Price on requestConfiguration and region to be confirmed
GLOBALATLAS AI · H200
NVIDIA H200 · 141 GB
141 GB is GPU memory, not system RAM. Confirm the number of GPUs and required interconnect before choosing a training topology.
Hopper141GB VRAM
- Memory technology
- HBM3e
- Workloads
- Memory-intensive LLM inference and scientific computing
More room for weights and context
- 01
141 GB HBM3e reference memory
- 02
Combined model and KV-cache budget
- 03
Exact GPU count and serving topology
H200 is a candidate when an H100 memory budget is restrictive. Its larger VRAM does not automatically establish more requests per second.
Regional configurations
USA
Availability on request
Price on requestConfiguration and region to be confirmed
Asia
Availability on request
Price on requestConfiguration and region to be confirmed
Choosing H200
When is the extra memory relevant?
When model weights, context cache and working buffers cannot fit the smaller configuration. Measure the complete memory peak before deciding to move from H100.
Technical references
Specification 1 →YOUR CONFIGURATION
Request an infrastructure quote
Choose resources and operating time. We will confirm a price for your configuration.
Each resource has its own listed configuration. Multiple GPU servers do not form an NVLink cluster. Monthly rates cover the full billing period; storage and traffic outside the configuration are not included.