Instances›AWS EC2›trn2›trn2.48xlarge
Compute
vCPUs192
Memory2048 GiB
Mem / vCPU10.67 GiB
ProcessorXeon Scalable
Clock speedUp to 3.9 GHz
Architecturex86_64
Storage
Type4 × 1,900 GB NVMe SSD
Bandwidth80,000 Mbps
Throughput10000 MB/s
IOPS260,000
Networking
Performance16x 200 Gigabit
Max ENIs32
IPv6Yes
GPUNone
Family sizes
trn2.48xlarge192 vCPU · 2048 GiB
trn2.48xlarge

trn2.48xlarge is AWS's machine-learning accelerator EC2 size at 192 vCPU and 2048 GiB. In the name, "trn" is Trainium, AWS's training silicon, "2" is the generation and "48xlarge" is the size within the trn2 family. It runs on Intel Xeon Scalable (Sapphire Rapids) at Up to 3.9 GHz, and AWS introduced it in December 2024. trn2 is the current generation of the trn family, so there is no newer AWS size to migrate to. Rackspace Spot has no equivalent to offer: the Inferentia and Trainium accelerators are AWS-exclusive silicon with no third-party equivalent. Spot's largest VM tops out at 16 vCPU and 128 GiB, and even that is not the missing piece here. The table below prices what AWS charges; the honest answer on the Spot side is that this workload stays where it is.

Match qualityNo equivalent

trn2.48xlarge is Machine Learning ASIC hardware. That accelerator has no counterpart in the Rackspace Spot catalog, and nothing in it emulates one at a speed worth paying for — this is a genuine no-match, not a weak one.

Rackspace Spot sells x86_64 Kubernetes nodes up to 16 vCPU, plus one NVIDIA A30 GPU class. What does move is the CPU half of the pipeline built around the accelerator: ingest, pre- and post-processing, orchestration, and the serving front end. On most accelerator workloads that tier runs continuously while the accelerator itself is duty-cycled, which is where the bill actually accumulates.

Pricing by region
1 hr
RegionAWS on-demandAWS Spot1-yr Reserved3-yr ReservedRackspace SpotYou save
US East, N. Virginia
US Central, Chicagono AWS region
US Central, Dallasno AWS region
Hong Kong
Sydney, Australia
London, UK
US West, San Jose
US Central, Dallas 2no AWS region
US East, N. Virginia 2
When to use it
Good for
  • ●Cost-per-inference optimisation inside AWS
  • ●Inference or training on AWS's own accelerator silicon
  • ●Workloads already built against the Neuron SDK
  • ●Workloads that fit inside 2048 GiB of RAM
  • ●Jobs that want 4 × 1,900 GB of local NVMe scratch space
  • ●Network-heavy services — AWS rates this size at 16x 200 Gigabit
Avoid if
  • ●The working set needs more than 2048 GiB of RAM
  • ●GPU or FPGA acceleration is needed
  • ●Frameworks that have not been compiled for the Neuron SDK.
  • ●More than 10.67 GiB of RAM per vCPU is needed — that ratio is fixed for the whole family
  • ●The work is single-threaded. Most of these 192 vCPU would sit idle.
Rackspace Spot sells spare capacity at market prices. You pay what the market decides, not a list price.
  • Spot sells spare capacity at market prices. It is the cheapest compute on the internet.
  • Prices are set by supply and demand. You pay what the market decides.
  • Compute is delivered as fully managed Kubernetes. All infrastructure is taken care of.
  • Egress costs $0.01/GB after the free allowance. AWS charges $0.09/GB.
Build your Cloudspace
Frequently asked questions