AMD Instinct MI430X

The AMD Instinct MI430X is an MI400-series GPU AMD built for sovereign AI and scientific computing.

Vendor
AMD
Architecture
Instinct MI400 series
Memory
432 GB HBM4
Memory bandwidth (TB/s)
23.3 TB/s
FP4 (PFLOPS)
9.2 PFLOPS
Form factor
GPU for HPC and sovereign AI systems
Status
Announced

The MI430X pairs up to 288 TFLOPS of hardware FP64 with 432 GB of HBM4 at 23.3 TB/s, and AMD gives up to 9.2 PFLOPS peak theoretical FP4/MXFP4. AMD launched it on July 23, 2026 and says it is expected to be available in 2027. AMD names the US DOE/ORNL Discovery system and Europe’s Alice Recoque and Herder systems as planned deployments.

Memory (GB)
432 GB
Precision note (sparse or dense)
AMD gives up to 9.2 PFLOPS peak theoretical FP4/MXFP4 and 288 TFLOPS hardware FP64, with no sparsity label.
Launch
Launched July 23, 2026; expected availability 2027
Vendor source
amd.com
Checked

Questions

When will the AMD MI430X be available?
AMD launched the MI430X on July 23, 2026 and says it is expected to be available in 2027.
How much memory does the MI430X have?
The MI430X has 432 GB of HBM4 with 23.3 TB/s of memory bandwidth.
What is the MI430X’s FP64 performance?
AMD lists up to 288 TFLOPS of hardware FP64, alongside up to 9.2 PFLOPS of FP4/MXFP4.
Which systems will use the MI430X?
AMD names the US DOE/ORNL Discovery system and Europe’s Alice Recoque and Herder systems as planned deployments.

Source

Every figure is taken from the vendor’s own datasheet, product page, technical documentation or announcement, linked on each chip with the date it was last checked. How we pick and check sources: editorial standards.

The Model Press

What are you looking for?

Search by headline, topic or keyword.