Parts/Accelerators/AWS Trainium2
Accelerators · Trainium · 2024
AWS Trainium2
Amazon's second-generation in-house training accelerator, deployed in UltraServer and UltraCluster topologies.
Why this part
The same logic as Graviton applied to accelerators: at Amazon's scale, designing the part is cheaper than paying someone else's margin on it, and the roadmap then matches the fleet. Trainium completes an AWS-designed node, with Graviton as host and Elastic Fabric Adapter as the network, none of which is purchasable as hardware. A growing share of the world's compute now exists only as rentable capacity, and this dataset can see it only where the operator chooses to publish.
Systems using this part
- Systems
- 4
- Rmax underneath
- 0.0 GFlop/s
- Units recorded
- 587,864
| System | Role | Country | Units | System Rmax | Supplied by | Rank |
|---|---|---|---|---|---|---|
| AWS Ridgeland Campus 2026-05 | Accelerator | USA | 261,200 | - | Amazon | - |
| Project Rainier 2025-10 | Accelerator | USA | - | - | Amazon | - |
| AWS Madison Mega Site (Canton) 2025-06 | Accelerator | USA | 326,600 | - | Amazon | - |
| AWS Trainium2 UltraServer | Accelerator | USA | 64 | - | Amazon | - |
“Units recorded” sums only the edges where a public source states a quantity. It is a floor, not a total, and should never be read as installed base.
Related news
- AWS activates Project Rainier, a nearly 500,000-chip Trainium2 cluster for Anthropic · About AmazonRainier is the largest non-NVIDIA training deployment in this dataset by chip count and the clearest evidence yet that a hyperscaler's own silicon, not merchant GPUs, can anchor a frontier model training run. AWS says Anthropic is headed toward more than 1 million Trainium2 chips by year end, which would make this entry's chip count stale within months rather than years.