Uploaded May 2023 | Updated September 2026, 2 weeks ago
The fastest AI appliances, be they NVIDIA DGX, HGX or EGX, require huge amounts of data, delivered at a consistently high speed to maximise GPU utilisation, getting the best ROI.
Unlike corporate networks which transmit data at relatively low speeds (10Gbs Ethernet), an efficient AI infrastructure would typically run at 100Gbs InfiniBand and faster, requiring specialist network cards, switches and interconnects. Larger deployments will also take advantage of DPUs to further optimise data flow.
Find out more scan.co.uk/business/ai-ecosystems/fast-connectivity
Get in touch with the Scan AI team to find out more - ai@scan.co.uk / 01204 474747
Follow us on social media:
LinkedIn - linkedin.com/company/scan-computers-international-ltd-
Facebook - facebook.com/ScanBusiness
Twitter - twitter.com/scan_business
The fastest AI appliances, be they NVIDIA DGX, HGX or EGX, require huge amounts of data, delivered at a consistently high speed to maximise GPU utilisation, getting the best ROI.
Unlike corporate networks which transmit data at relatively low speeds (10Gbs Ethernet), an efficient AI infrastructure would typically run at 100Gbs InfiniBand and faster, requiring specialist network cards, switches and interconnects. Larger deployments will also take advantage of DPUs to further optimise data flow.
Find out more scan.co.uk/business/ai-ecosystems/fast-connectivity
Get in touch with the Scan AI team to find out more - ai@scan.co.uk / 01204 474747
Follow us on social media:
LinkedIn - linkedin.com/company/scan-computers-international-ltd-
Facebook - facebook.com/ScanBusiness
Twitter - twitter.com/scan_business










