Compare instance

trn2.48xlarge

⚡ 16x GPU • 1536.0 GB HBM3

Part of the TRN2 Accelerated Computing family — Hardware accelerators (GPUs, FPGAs) for specialized compute.

Name:trnAWS Trainium Acceleratorstrn = AWS Trainium AcceleratorsPurpose-built AWS Trainium silicon engineered for high-performance, cost-effective deep learning model training.22nd Generation2 = 2nd GenerationAWS 2nd generation hardware platform with updated CPU/system architecture.48xlarge48 Extra Large48xlarge = 48 Extra Large192 vCPUs • 2,048 GB RAM • EBS only • 3200 Gbps Network • 16x GPU (1536.0 GB VRAM)
Price (Linux)
$103.1570/ hr
~$75,304.61/mo
Compute
192vCPUs
96 cores • x86_64
Memory
2,048GB
10.7 GB / vCPU
Value
$0.5373/ vCPU-hr
$0.0504/GB-hr
Network
3200 Gbps
Baseline: 200 Gbps
Storage & EBS
10,000MB/s
Max: 260,000 IOPS

Technical Specifications & Limits

Hardware specs, container thresholds, and system limits

Compute & Processor

ProcessorIntel Xeon Scalable (Sapphire Rapids)
Architecturex86_64 (x86_64)
vCPUs / Threads192 vCPUs
Physical Cores96
Clock SpeedUp to 3.9 GHz
Intel AVX / AVX2AVX & AVX2 Supported
Intel Turbo BoostYes
Bare MetalNo (Virtualized)

Performance & Benchmarks

CoreMark ScoreNot benchmarked
CoreMark / vCPU—
FFmpeg Transcoding—
CoreMark per Dollar—
Cost per vCPU-hr$0.5373
Cost per GB-hr$0.0504

Containers & Kubernetes

EKS Max Pods100 pods
Max ECS Tasks—
Max ENIs32
IPv4/v6 per ENI50
Total Addressable IPs1600 total IPs
CNI Prefix DelegationSupported (High Density)

Networking & Bandwidth

Network Performance3200 Gbps
Baseline Bandwidth200 Gbps
Burst / Peak Bandwidth200 Gbps
Enhanced NetworkingYes
EFA SupportNo
IPv6 SupportNative Dual-Stack Supported

Storage & EBS Optimization

Storage ConfigurationEBS only
Local NVMe DiskNone (EBS only)
EBS Baseline Throughput10000 MB/s
EBS Max Throughput10000 MB/s
EBS Baseline IOPS260,000
EBS Max / Burst IOPS260,000
Dedicated EBS Throughput80 Gbps

GPU & Hardware Acceleration

GPU ModelHBM3
Accelerator Count16x GPU(s)
Total GPU Memory1536.0 GB
Memory ArchitectureHBM3
CUDA Compute Capability—
Acceleration RuntimesCUDA, TensorRT, cuDNN, PyTorch, PyJAX

Platform & Architecture Details

Hypervisor / SystemAWS Nitro System (PCIe Offload)
Nitro EnclavesSupported (Isolated security enclaves)
Normalization Factor384
Generation StatusCurrent Generation (Recommended)
Date Introduced2024-12-03

Regional Pricing Matrix

Live on-demand rates across all global AWS cloud regions
Lowest Price Region
$103.1570/ hr
Asia Pacific (Hyderabad) (ap-south-2)
Highest Price Region
$103.1570/ hr
Asia Pacific (Hyderabad) (ap-south-2)
Global Fleet Average
$103.1570/ hr
Across all active regions
Regional Coverage
1regions
Available worldwide
Linux — Available in all 1 regions
RegionOn-Demand Price
ap-south-2
Asia Pacific (Hyderabad)
Lowest
$103.1570