The available history contains 22 observations from 2026-05-19 to 2026-10-20. These are recorded dates, not proof of the earliest available publication; missing intervals are not interpolated.
Nine leading cloud providers are now seen spending over $1.5 trillion in 2027, with AI server revenue rising more than 90% on rack-level shortages and higher memory quotations.
Intel's first Xe3P data-center part trades peak compute for memory capacity: 1.3 PFLOPS of FP8, 1.5 TB/s of bandwidth, 350W and up to 480GB of LPDDR5X in a standard PCIe card aimed at agentic AI.
Microsoft and Intel unveiled in-house AI accelerators with opposite memory strategies at the OCP Global Summit: Maia 300 moves to eight HBM4 stacks on TSMC 2nm, while Intel's new GPU D-Series uses up to 480GB of cheaper LPDDR5X.