Dubbed Vela, the AI has been running since May last year, and is touted as IBM's first AI-optimized, cloud-native supercomputer. It has been built to create and training large-scale AI models. 

Biggish Blue said that the platform is currently reserved for use by the IBM Research community and has become popular with researchers creating advanced AI capabilities since May 2022, including work on foundation models.

IBM states that it chose this architecture because it gives the company greater flexibility to scale up as required, and also the ability to deploy similar infrastructure into any IBM Cloud datacenter around the globe.

Vela is based around a twin-socket system with 2nd Gen Xeon Scalable processors configured with 1.5TB of DRAM, and four 3.2TB NVMe flash drives, plus eight 80GB Nvidia A100 GPUs, the latter connected by NVLink and NVSwitch.

This makes the Vela infrastructure closer to that of a high performance compute site than typical cloud infrastructure, despite IBM's insistence that it was taking a different path as "traditional supercomputers weren't designed for AI."

IBM used x86 processors rather than its own Power 10 chips which is a little odd as large-model AI inferencing was what the Power 10s were designed for.