Intel has reportedly licensed register-transfer level code for Atom-class x86 CPU technology to newly formed startup RosaicLabs, an unusually deep handoff that could let the company develop custom processors compatible with the PC industry’s dominant instruction set. As first reported by Reuters and detailed by TechSpot, the agreement is notable not because it changes any current Windows PC, but because Intel has rarely made this level of CPU implementation IP available outside its own operations. RosaicLabs was incorporated in Delaware only recently, according to the reporting, and is led by Amarjit Gill, a longtime business associate of Intel CEO Lip-Bu Tan. Reuters also identified former Rivos finance executive Amit Parikh as part of the venture. Neither the specific Atom generation nor the licensing terms have been disclosed publicly.

A high-tech lab blends chip design, code, robotics, drones, and cleanroom manufacturing.This Is More Than an x86 Compatibility License​

The reported asset is Atom RTL—the hardware-description code used to model how data moves between registers and logic blocks before a processor is physically implemented. That is materially different from merely receiving permission to execute or support x86 instructions.
If the report is accurate, RosaicLabs could adapt an existing low-power Intel CPU design, integrate it with specialized accelerators or I/O, and turn it into a custom chip. The deal does not necessarily mean RosaicLabs can freely sell a general-purpose rival to Intel Core or Xeon processors; licensing restrictions, patent coverage, manufacturing choices, validation requirements and software support could all sharply limit what it can build.
Still, this is a departure from Intel’s traditional posture. AMD’s long-running x86 cross-license remains the best-known example of an external relationship around x86, but Intel has not operated as a broad merchant supplier of CPU-core IP in the mold of Arm or RISC-V vendors.

Atom’s Value May Be at the Edge, Not in a New PC​

The Atom name is often associated with netbooks, but Intel’s small-core designs have continued in embedded, industrial and low-power product lines—and their descendants underpin the efficiency cores used in modern Intel hybrid processors. A startup using Atom-derived RTL would therefore have a plausible path toward fanless edge hardware, industrial controllers, robotics systems, networking appliances or tightly integrated AI inference devices.
For Windows users and IT administrators, the attraction is straightforward: a specialized x86 system can retain conventional Windows and Linux software compatibility without the translation and driver-transition questions that accompany Arm-based deployments. That does not guarantee Windows support for any eventual RosaicLabs silicon; Windows certification, firmware quality, graphics, networking and security features matter as much as instruction-set compatibility.
The immediate effect on the PC market is therefore close to zero. There is no announced RosaicLabs processor, platform, foundry partner, launch date or Windows device.

Intel May Be Testing a Different Use for Its CPU IP​

The larger signal is strategic. Intel has spent years trying to expand Intel Foundry into a business that serves outside chip designers, and a selectively licensed Atom-derived core could give a customer a familiar x86 base while creating a reason to manufacture a differentiated chip through Intel’s ecosystem.
That reading remains inference rather than a disclosed product plan. Reuters’ account does not establish whether RosaicLabs must use Intel manufacturing, whether it has rights to modify the design extensively, or whether Intel retains options to buy the company or its resulting technology.
For now, the story is a rare indication that Intel may see older or specialized x86 CPU designs as assets it can license into markets it does not intend to serve directly. The next meaningful milestone will be evidence of what RosaicLabs is actually building—and whether its first chip is a conventional embedded processor or something far more tailored to AI and robotics.

References​

  1. Primary source: TechSpot
    Published: 2026-07-31T18:53:00+00:00
  2. Related coverage: tomshardware.com