Intel (INTC) accelerates open infrastructure for the agentic AI era; begins customer sampling of GPU d-series

Product and sampling announcements from Intel have historically drawn a muted tape response relative to peers in the accelerator space, reflecting a track record in which announced data-centre silicon roadmaps have repeatedly slipped between sampling and volume revenue. The operative distinction is between customer sampling and production ramp: sampling is the early stage of a qualification cycle that in this industry typically runs many quarters, and revenue attribution in the segment disclosures has been the tell that separates real traction from press-release momentum. The peer-set read-through matters here, since the dominant AI accelerator vendor's share position and the custom-silicon programs at the large cloud buyers define the bar Intel's d-series must clear, and design-win commentary from hyperscalers has historically carried more signal than the vendor's own framing. Worth noting is that open-infrastructure positioning is a deliberate contrast with the incumbent's proprietary stack, a strategy challengers in this market have pursued before with mixed results. The follow-ons are management commentary on the next earnings call, any named-customer confirmations, and foundry versus product segmentation in the reporting.