AMD MI355X

AMD MI355X

AMD's next-gen CDNA 4 competitor to Nvidia Blackwell.

Compare vs other GPUs →
Aggregating historical prices...

Key Specifications

Architecture
CDNA
Memory
2.3TB HBM3E
Memory Bandwidth
8,000 GB/s
Release date
Q2 2025

Compare Cloud Provider Prices

We're tracking 5 cloud providers for the MI355X, and the pricing spread is significant. While the average sits at $5.45/hr, the lowest price is currently $2.29/hr per GPU (36-month reservation).

of shown

Default ranking

Our algorithm weighs five factors to find the most relevant matches for you:

  1. Price: We blend total hourly price with price-per-GPU to balance affordability and value.
  2. Specs: We favor offers with higher CPU, RAM, and GPU Memory.
  3. Billing: We favor on-demand billing for simplicity and flexibility over spot instances, reservations, and custom quotes.
  4. Location: We blend both datacenter proximity and provider HQ location. Datacenter location matters for latency, while HQ location matters for compliance and support.
  5. Provider diversity: Each time a provider appears in the list, their subsequent offerings are ranked a little lower, so one provider's offerings don't crowd out the top positions.

Sorting and filtering

Click any column header to sort by that column. Use the filters above the table to narrow results by billing type, GPU count, vCPUs, or RAM. Custom sorting resets the default relevance ranking.

Transparency and funding

Ads and sponsors: Any paid placements are fixed at the top and clearly labeled as sponsored content.

Affiliates: Any affiliate links will be indicated to you as well. We may earn a commission if you click them, but this never influences the ranking order.

Provider GPUs Total VRAM vCPUs RAM Billing $/GPU/h Total/h Availability
Vultr logo Vultr 8x MI355X 8x MI355X 288GB 2.3TB 128 3TB Reserved (36mo) Reserved capacity with a 3-year minimum commitment. $2.29 $18.32 Unknown View
Oracle Cloud logo Oracle Cloud 8x MI355X 8x MI355X 288GB (BM.GPU.MI355X.8) 2.3TB -- -- On-Demand Pay-as-you-go pricing. No term commitments. $8.60 $68.80 Unknown View
TensorWave logo TensorWave 8x MI355X 8x MI355X 2300GB 18.4TB -- -- Custom Custom terms negotiated with the provider. Custom Custom Unknown View
Hot Aisle logo Hot Aisle 8x MI355X 8x MI355X -- -- -- Custom Accepting reservations Custom Custom Waitlist View
Crusoe logo Crusoe 8x MI355X 8x MI355X 288GB (MI355X) 2.3TB -- -- On-Demand Pay-as-you-go pricing. No term commitments. Custom Custom Unknown View
No offerings matching your filters.

Heads up: We do our best to keep these prices accurate. However, cloud costs may fluctuate based on region, usage, and other factors not listed here. These are estimates based on common setups and are for informational purposes only. Always verify current rates with the provider before provisioning.

Frequently Asked Questions

Why choose the MI355X?

288GB HBM3e with FP4 support on CDNA 4 architecture. Next-generation AMD compute for both training and inference. Competitive with Nvidia Blackwell on memory capacity.

When is the MI355X not a good fit?

Latest generation with minimal cloud availability. ROCm software stack, while rapidly improving, may lag behind CUDA for some workloads. Evaluate software compatibility before committing.

Are MI355X prices going up or down?

On-demand pricing has increased by about 232% since October 2025, from $2.59 to $8.60/hr per GPU.

What size AI models can the MI355X run?

With 2300GB of VRAM, the MI355X can usually run 70B-class models with headroom, and may handle much larger models in 4-bit quantized form depending on runtime overhead, KV cache, and context length.

How much VRAM does the MI355X have?

The MI355X has 2300GB of VRAM. Multi-GPU setups increase total memory, but that memory is not automatically pooled across GPUs.

What is the MI355X's memory bandwidth?

The MI355X has 8,000 GB/s of memory bandwidth. Higher bandwidth helps with faster data transfer between GPU memory and compute cores.

What data types does the MI355X support?

The MI355X supports 7 precision formats. Training: BF16, FP16, FP32. Inference: FP4, FP8, INT8. Scientific: FP64.

Does the MI355X support Infinity Fabric?

Yes. The MI355X supports Infinity Fabric with 896 GB/s of bidirectional bandwidth. This helps accelerate multi-GPU communication.

How much does the MI355X cost per hour?

MI355X pricing currently ranges from $2.29/hr to $8.60/hr per GPU, depending on the provider, instance type, and billing model.

How much does the MI355X cost per month?

At 720 hours per month, one MI355X can cost between $1,648.80 to $6,192.00 per month, depending on the provider. Reserved and spot pricing can lower that further.

Which cloud providers offer the MI355X?

The MI355X is available from 5 cloud providers, including Crusoe, TensorWave, Vultr. Pricing and availability vary by region and billing model.

Can I rent the MI355X in the cloud?

Yes. We currently track 5 MI355X listings across 5 cloud providers:

Billing type Listings Avg $/GPU/hr
On-demand 2 $8.60/hr
Reserved 1 $2.29/hr
Custom contract 2 Custom

Technical Specifications

FP16 Performance 20.1 PFLOPS (w/ sparsity: 40.3 PFLOPS)
BFLOAT16 Performance 20.1 PFLOPS (w/ sparsity: 40.3 PFLOPS)
INT8 Performance 40.3 POPS (w/ sparsity: 80.5 POPS)
INT4 Performance 40.3 POPS (w/ sparsity: 80.5 POPS)
FP8 Performance 40.3 PFLOPS (w/ sparsity: 80.5 PFLOPS)
FP6 Performance 80.5 PFLOPS (w/ sparsity: 161.1 PFLOPS)
FP4 Performance 80.5 PFLOPS (w/ sparsity: 161.1 PFLOPS)
FP64 Vector Performance 628.8 TFLOPS
FP32 Vector Performance 1.3 PFLOPS
FP64 Matrix Performance 628.8 TFLOPS
FP32 Matrix Performance 1.3 PFLOPS
Memory Capacity 2.3 TB HBM3E
Memory Bandwidth 8 TB/s max. peak theoretical
Memory Interface 8192 bits
AMD Infinity Cache™ (Last Level) 256 MB (per GPU)
Form Factor UBB 2.0 module
Lithography TSMC 3nm/6nm FinFET
GPU Compute Units 2048
Matrix Cores 8192
Stream Processors 131,072
Peak Engine Clock 2200 MHz
Scale-up AMD Infinity Fabric™ Links 7 (per GPU) x 153.6 GB/s
I/O to Host CPU 8 PCIe® Gen 5 x16 (128 GB/s)
Video Decoders 32 groups for HEVC/H.265, AVC/H.264, V1, or AV1
JPEG/MJPEG CODEC 320 cores, 10 cores per group
GPU Physical Partitions SR-IOV, up to 64 partitions
Memory Partitions 1 or 4 per module
Coolant PG-25
Maximum Liquid Inlet Temperature 43°C
Recommended Flow Rate 2.1 l/min per OAM
RAS Features Full-chip ECC memory, page retirement, page avoidance
Maximum TBP 1400W per module

Source: official AMD MI355X datasheet.

Alternatives to AMD MI355X

Last updated