Intel’s 4th Gen Xeon Scalable processors have once again been delayed due to a slower ramp despite increased demand from all sides. Although the official state of the Sapphire Rapids family is “launched”, most customers are yet to receive retail units. This has likely pushed many small-scaled firms towards alternatives, including Ice Lake-SP or AMD’s Milan/Milan-X offerings.
NVIDIA recently picked Sapphire Rapids as the companion chip to its H100 data center GPU. Intel has confirmed this, admitting that the GPU has the unique position of being both a customer and a competitor. Earlier, the G100 had ditched Intel CPUs for the first time to opt for AMD’s Epyc Milan chip on account of its superior I/O technologies.
Despite this delay, the chipmaker believes that its 7nm process (formerly 10nm ESF) is one of its healthiest nodes ever. Citing the successful shipment of 15 million Alder Lake units is a neat feat, it’s not the same as shipping SKUs with four 15 core chiplets, all fully functional at the top-end.