Do you need a high-performance accelerated air-cooled server optimized for large AI models? The HPE Compute XD690 extends the NVIDIA AI Computing by HPE portfolio and is designed for the escalating requirements of large AI model training, tuning and inferencing. Featuring eight NVIDIA Blackwell Ultra (B300 HGX) GPUs, delivering 50% higher FP4 performance and double the GPU-CPU bandwidth of the prior Blackwell model ^[1], it can support highly complex models and AI reasoning workloads at massive scale. HPE delivers a broad range of services to speed up the deployment of HPE Compute XD690 systems and mitigate risk. Unlike other vendors, HPE has decades of experience delivering and managing the largest, most complex systems. We bring that expertise to you as you deploy a large AI environment powered by the HPE Compute XD690. Combined with HPE’s industry-leading services and support, this server delivers exceptional performance and scalability to handle the largest AI models.

What is new supporting point:
8x NVIDIA Blackwell Ultra (B300 HGX)
2x Intel Xeon 6 processors
10U, air-cooled
Support for 96 GB DIMMs
HPE InfiniBand XDR and NVIDIA Spectrum-X Ethernet support
OpenBMC and Redfish APIs allow for industry standard BMC control of the server