Compare instance

AWS Amazon EC2

p6-b300.48xlarge

⚡ 8x GPU • 2144.0 GB HBM3e

Part of the P6-B300 Accelerated Computing family — Hardware accelerators (GPUs, FPGAs) for specialized compute.

Name:pGPU Accelerated Computingp = GPU Accelerated ComputingHigh-end NVIDIA GPUs (Blackwell, Hopper, Ampere, Volta) built for frontier AI/LLM training, HPC simulations, and generative AI.66th Generation6 = 6th GenerationAWS 6th generation GPU platform.b300NVIDIA Blackwell B300b300 = NVIDIA Blackwell B300Intel Xeon Scalable (Emerald Rapids) (x86_64). Accelerated by high-performance NVIDIA Blackwell B300 GPUs with NVLink interconnects for frontier deep learning models.48xlarge48 Extra Large48xlarge = 48 Extra Large192 vCPUs • 4,096 GB RAM • 30720 GB NVMe SSD • 6400 Gbps Network • 8x NVIDIA B300 (2144.0 GB VRAM)
On-demand, Linux
$142.4160/ hr
~$103,963.68/mo · lowest in US East (Atlanta), us-east-1-atl-1
Savings Plan, Linux
$104.59031/ hr
From, with a 3-year Compute Savings Plan, No Upfront: −27% vs on-demand rate (us-east-1-atl-1)
Per unit
$0.7418/ vCPU-hr
$0.0348/GB-hr
Compute
192vCPUs
96 cores • x86_64
Memory
4,096GB
21.3 GB / vCPU
Network
6400 Gbps
Baseline: 6400 Gbps
Storage & EBS
12,500MB/s
Max: 400,000 IOPS

On Google Cloud

p6-b300.48xlarge is an accelerated machine. Those are compared by the accelerator, not by vCPU and memory, so this page doesn't pick a match. Browse Google Cloud machines.

Technical Specifications & Limits

Hardware specs, container thresholds, and system limits

Compute & Processor

ProcessorIntel Xeon Scalable (Emerald Rapids)
Architecturex86_64 (x86_64)
vCPUs / Threads192 vCPUs
Physical Cores96
Clock Speed3.2 GHz
Intel AVX / AVX2No
Intel Turbo BoostYes
Bare MetalNo (Virtualized)

Performance & Benchmarks

CoreMark ScoreNot benchmarked
CoreMark / vCPU—
FFmpeg Transcoding—
CoreMark per Dollar—
Cost per vCPU-hr$0.7418
Cost per GB-hr$0.0348

Containers & Kubernetes

EKS Max Pods198 pods
Max ENIs68
IPv4/v6 per ENI50
Total Addressable IPs3400 total IPs
CNI Prefix DelegationSupported (High Density)

Networking & Bandwidth

Network Performance6400 Gbps
Baseline Bandwidth6400 Gbps
Burst / Peak Bandwidth6400 Gbps
Enhanced NetworkingYes
EFA SupportNo
IPv6 SupportNative Dual-Stack Supported

Storage & EBS Optimization

Storage ConfigurationNVMe SSD (30720 GB)
Local NVMe Disk30,720 GB NVMe SSD
EBS Baseline Throughput12500 MB/s
EBS Max Throughput12500 MB/s
EBS Baseline IOPS400,000
EBS Max / Burst IOPS400,000
Dedicated EBS Throughput100 Gbps

GPU & Hardware Acceleration

GPU ModelNVIDIA B300
Accelerator Count8x GPU(s)
Total GPU Memory2144.0 GB
Memory ArchitectureHBM3e
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)

Rates by region

Live on-demand rates across all global AWS cloud regions
Lowest Price Region
$142.4160/ hr
US East (N. Virginia) (us-east-1)
Highest Price Region
$281.1300/ hr
South America (Sao Paulo) (sa-east-1)
Global Fleet Average
$190.8628/ hr
Across all active regions
Regional Coverage
10regions
Available worldwide
Linux — Available in all 10 regions
RegionOn-Demand Price
ap-south-2
Asia Pacific (Hyderabad)
$200.8066
ap-southeast-3
Asia Pacific (Jakarta)
$230.9300
ap-northeast-2
Asia Pacific (Seoul)
$209.1735
us-gov-east-1
AWS GovCloud (US-East)
$170.9000
us-gov-west-1
AWS GovCloud (US-West)
$170.9000
eu-west-2
EU (London)
$217.5400
sa-east-1
South America (Sao Paulo)
Highest
$281.1300
us-east-1-atl-1
US East (Atlanta)
Lowest
$142.4160
us-east-1
US East (N. Virginia)
Lowest
$142.4160
us-west-2
US West (Oregon)
Lowest
$142.4160