Leaked Intel SoC Plan Pairs x86 CPUs With NVIDIA RTX Integrated GPUs In Early 2028 - HotHardware

Leaked details suggest that Intel and NVIDIA are collaborating on a new system-on-chip (SoC) that integrates NVIDIA's RTX GPU chiplets with Intel's x86 CPUs. Expected to launch in early 2028 under the... Read original →

If Intel and NVIDIA launch an x86 SoC integrating RTX GPU chiplets by early 2028, it will trigger three technical ripple effects: first, heterogeneous chiplet integration will intensify demand for CoWoS-L and Foveros Direct packaging; second, CUDA may embed deeply into x86 client platforms for the first time, disrupting Apple’s Metal and AMD’s ROCm in local AI inference; third, Intel’s pivot from its Xe graphics to NVIDIA IP reveals persistent GPU architecture weaknesses. On compliance, U.S. BIS export controls—especially if NVIDIA entrusts Intel with future GPU fabrication—add significant geopolitical risk. Competitively, AMD may accelerate Zen5+RDNA4 unified memory designs, while Qualcomm leverages its ARM+Adreno stack in AI PCs. Over the next 12–24 months, this partnership could normalize ‘CPU + discrete-GPU IP’ SoCs—but without strong end-market pull, it risks repeating the Larrabee-era integration failure.