技术档案
AI推理算力AI inference compute
运行模型生成输出的算力
每条都摘自下方编号的来源原文；只写事实，不写预测或建议。
已有应用
- Anthropic与Amazon推出Project Rainier，并使用超过一百万颗Trainium2芯片训练和运行Claude。2
Together we launched Project Rainier, one of the largest compute clusters in the world, and we currently use over one million Trainium2 chips to train and serve Claude.
- AWS为OpenAI构建的集群既支持ChatGPT推理，也支持下一代模型训练。1
The clusters are designed to support various workloads, from serving inference for ChatGPT to training next generation models, with the flexibility to adapt to OpenAI's evolving needs.
为何此时
- Anthropic称2026年企业及开发者对Claude需求加速，同时消费者使用量显著上升。2
Enterprise and developer demand for Claude has accelerated in 2026, and alongside it we have experienced a sharp rise in consumer usage across our free, Pro, and Max tiers.
主要参与者
- AWS的Amazon EC2 UltraServers把NVIDIA GB200及GB300 GPU集群接入同一网络，为OpenAI运行工作负载。1
Clustering the NVIDIA GPUs—both GB200s and GB300s—via Amazon EC2 UltraServers on the same network enables low-latency performance across interconnected systems, allowing OpenAI to efficiently run workloads with optimal performance.