AWS Inferentia2
An inference-focused AWS chip with two NeuronCore-v2 cores. Inf2 instances can contain up to twelve chips.
HBM
Published peak, not measured application throughput
AWS documents 32 GiB and 820 GiB/s per chip. Twelve chips do not create one automatically usable allocation: the model must be partitioned across the selected Neuron topology.
When this is a sensible choice
Start here if…
Evaluate a supported Neuron model when inference is a sustained workload. Start with compilation and an accuracy check before comparing cost at your required latency.
Choose another configuration if…
A CUDA container cannot run here unchanged. A model that fits in HBM can still require unsupported operators or an unavailable parallel layout.
Specifications with their boundaries attached
- Architecture
- NeuronCore-v2
- Memory
- 32 GiB HBM
- Memory bandwidth
- 0.88 TB/s per accelerator
- Peak compute
- 190 TFLOPS BF16/FP16 per chip; 380 INT8 TOPS
- Scale-up interconnect
- NeuronLink-v2 between chips; instance configuration determines the domain
- Host attachment
- AWS-managed host; verify the selected EC2 configuration
- Power
- Not published in checked per-chip reference
- Partitioning
- Physical and logical NeuronCore allocation; not NVIDIA MIG
- Catalogue status
- Documented product
Compute figures are theoretical peaks at the stated precision. Dense and structured-sparse rates must not be mixed. Bandwidth labelled bidirectional combines both directions. See the source documents.
Three bandwidths, three different jobs
NeuronLink-v2 between chips; instance configuration determines the domain
InfiniBand, RoCE, EFA or provider-specific transport
Where it appears in provider documentation
| Provider / machine | Network scope | What changes the decision |
|---|---|---|
AWS Inf2 instance family ↗ | NeuronLink-v2 between chips; verify host networking for the selected size | The architecture reference documents up to twelve chips per instance. Check model partitioning and the exact size before adding chip memory together. |
Model fit and software support
These publisher or serving-engine documents mention this hardware family. They have not been reproduced on our machines.
No model-specific recipe in our reviewed set certifies this exact hardware. The memory calculator can narrow candidates, but it cannot establish software support. Read the model register.
What is the memory floor?
Start with total parameters, then add the memory the workload needs. This arithmetic does not certify a serving configuration. All output sizes below are decimal GB.
Override device capacity with the memory exposed by your allocation, particularly for cloud B300 and partitioned devices. The starting 32 GB budget and 15% reserve are editable teaching assumptions. They are not measurements for the selected model. Mixed-precision tensors, quantization scales, vision encoders and draft models can increase the weight payload.
Raw weights only
320B × 8 bits ÷ 8
Budget per device
34.36 GB × (1 − 15%)
Arithmetic lower bound
round up ((weights + 32) ÷ budget)
This assumes perfectly balanced sharding. A result of three does not prove that a three-device parallel layout is supported. Check layer/expert divisibility, actual allocatable memory, precision kernels and the fabric before renting.
320B is the model card’s total; 18B active is not its storage size. The vLLM recipe reports about 306 GiB for the native FP8 checkpoint. Hopper requires BF16 KV for this model; the documented ROCm path targets gfx950, not every Instinct GPU. Read the model source ↗
For full training, also budget gradients, optimizer states, activations and communication buffers. The inference weight estimate above is insufficient.
Tokens per second, TTFT and TPOT are not certified for this hardware in our reference. Use the benchmark checklist to compare an exact model, software revision and workload.
Nearby memory capacities, different tradeoffs
These are comparison candidates selected by memory capacity, not performance rankings or drop-in replacements.
AWS Trainium2
AWS’s own training accelerator, also used for supported inference. It runs the Neuron stack rather than a CUDA binary.
AWS Trainium3
Eight NeuronCore-v4 cores and 144 GiB of HBM per chip. Its low-precision and memory upgrades are much larger than its dense BF16 compute increase over Trainium2.
Sources and review scope
Manufacturer and cloud documentation checked 2026-09-12. We reviewed specifications and the stated product boundaries; we did not run training or serving benchmarks on this device. Cloud memory and availability can differ by configuration.
- AWS Neuron Inferentia2 architecture ↗ · checked 2026-09-12
Cite this reference
AI Infra Interviews. AWS Inferentia2: specifications and workload fit. Reviewed . https://aiinfrainterviews.com/hardware/inferentia2
Include your access date when citing a changing specification. Link to the specification section for hardware figures or the explanation for a sizing or workload decision.
Research method and limits
We compile manufacturer specifications, cloud documentation, model cards and serving recipes. The reference preserves source units and distinguishes individual devices from nodes and racks. Conflicting figures and unknown fields remain labelled.
Our contribution is the comparison, unit reconciliation, worked arithmetic and workload explanation. Published peaks are vendor specifications. Calculator results are estimates under the displayed assumptions. Neither is a measurement from our own accelerator lab.
For a manufacturer’s specification, consult the original source documents. Cite our page when using its analysis, and retain primary-source attribution for underlying figures. This curated reference does not establish market share, live availability or a universal performance ranking.
Found a discrepancy? Send a correction with the page URL, exact variant, disputed figure and supporting primary document.