Data-center AI inference accelerator from HyperAccel
hyperaccel.ai/ha_product/bertha-500
Bertha 500 (B500) is HyperAccel's data-center class Latency-Processing Unit accelerator for LLM inference. Public product materials describe high batch concurrency, LPDDR5X memory, and a dual-slot PCIe form factor aimed at hyperscale and on-prem inference deployments. Product page: hyperaccel.ai/ha_product/bertha-500/.