NVIDIA Replaced Proprietary Cores With RISC-V

NVIDIA integrated RISC-V cores to manage internal GPU control tasks, replacing its legacy Falcon architecture.

Updated on Sept. 21, 2026 in Semiconductors

Bold flat-color editorial illustration showing a simplified geometric processor die, evoking the precision of semiconductor hardware architecture. — NVIDIA has migrated its internal GPU control tasks to RISC-V architecture, replacing the proprietary Falcon system to enable 64-bit addressing and complex OS management. AI Illustration. Upload story photo >

Live Poll

Do you prefer companies provide full transparency about the internal components used in your devices?

NVIDIA transitioned from its internal Falcon microcontroller architecture to RISC-V to handle management operations within its GPUs. This architectural shift, which has been in progress since at least 2018, allows for more robust control tasks outside of standard shader computation.

Why it matters

The transition to RISC-V provides NVIDIA with capabilities its legacy hardware lacked, specifically 64-bit addressing and the ability to run more complex operating system tasks. This migration underscores the industry-wide move toward open-standard instruction sets for embedded management operations.

Current NVIDIA GPUs incorporate between 10 and 40 RISC-V cores for internal management, significantly surpassing the capability of the retired Falcon architecture. Unlike Falcon, which lacked 64-bit addressing, the RISC-V implementation supports more complex control logic and firmware tasks.

The players

NVIDIA

A designer of graphics processors and hardware stacks known for high-performance computing and AI infrastructure.

The details

Embedded RISC-V cores function as system-management units, executing tasks that occur alongside the main graphics-processing hardware without interrupting the core shader pipeline. By moving to this architecture, NVIDIA bypassed the limitations of Falcon, a proprietary instruction set that was unable to run a full operating system or manage advanced memory addressing. These cores operate as black-box controllers, meaning end users cannot directly program or interface with them, keeping their operations entirely separate from user-facing driver code.

2018: A specialized core began handling graphics-driver execution tasks.

The Tech Race

NVIDIA’s move away from the proprietary Falcon microcontroller architecture represents a broader industry trend toward adopting RISC-V for embedded control. This shift places NVIDIA in line with other chip designers who are increasingly prioritizing open-standard architectures for internal firmware management.

This transition happens entirely behind the scenes, as these cores manage internal hardware tasks and are not user-programmable. Users do not need to take any action, as these firmware operations are invisible to standard GPU software workflows.

The takeaway

NVIDIA successfully phased out its legacy Falcon system in favor of RISC-V to modernize its internal control hardware. Watch for further technical disclosures on how these embedded cores are utilized in next-generation GPU firmware deployments.