🤖 AI Summary
Nvidia and Intel announced a strategic partnership in which they will co-develop multiple generations of x86 products and Nvidia will buy $5 billion of Intel stock at $23.28/share. The collaboration covers two headline items: "Intel x86 RTX SoCs" for consumer gaming PCs — combining an Intel x86 CPU chiplet tightly connected to an Nvidia RTX GPU chiplet in one package — and custom Intel‑fabricated x86 data‑center CPUs that Nvidia will sell to hyperscale and enterprise customers. Both sides promise multi‑generation roadmaps, but product specs and timelines are unspecified; realistic commercialization will likely take a year or more.
Technically, the partners emphasize much tighter CPU–GPU integration than traditional PCIe-linked systems by using NVLink (and Nvidia’s NVLink Fusion), promising up to ~14× more bandwidth and lower latency plus uniform memory access (UMA) so CPU and GPU can share the same memory pool. Manufacturing split is undecided: some data‑center silicon could be fabbed on Intel nodes (Intel 3, 18A), while client chips or GPU chiplets might still use TSMC. Nvidia says it remains committed to its other CPU initiatives (Grace, Grace Blackwell, Vera), and driver responsibilities will be split (Nvidia for GPUs, Intel for consumer processors). The deal reshapes x86/GPU dynamics, accelerates CPU–GPU convergence for laptops and servers, and has strategic financial/geopolitical weight given concurrent investments from the U.S. government and SoftBank.
Loading comments...
login to comment
loading comments...
no comments yet