Europe
Availability on request
Price on requestConfiguration and region to be confirmed
GLOBALATLAS AI · Instinct MI300X
AMD Instinct MI300X · 192 GB
MI300X has 192 GB HBM3 per accelerator. It uses the ROCm software stack; CUDA-only containers need a compatible build.
CDNA 3192GB VRAM
- Hardware variant
- OAM
- Memory technology
- HBM3
- Workloads
- LLM inference and ROCm training
Configuring Instinct MI300X
- 01
Check ROCm support in the model runtime
- 02
Budget weights, KV cache and temporary buffers within 192 GB
- 03
Specify OAM platform, accelerator count and Infinity Fabric topology
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
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.