Nvidia is enabling outside chip designers to build custom silicon that works with its systems, but the available announcements do not show a general program licensing Nvidia graphics cores to any chip vendor. The clearest example is NVLink Fusion, a way for custom XPUs to connect into Nvidia-based AI infrastructure. That matters commercially because a customer can design its own accelerator and still rely on Nvidia technology elsewhere in the system.
What Nvidia has actually said about licensing
Nvidia’s 2026 annual-review materials describe using its intellectual property and working with partners through license and development agreements when they want to build capabilities into their products. The company’s 2026 SEC filing also says it uses patents, trademarks, trade secrets, nondisclosure agreements and licensing arrangements to protect intellectual property, and expects to continue entering license agreements.
Those disclosures establish that licensing is part of Nvidia’s business practice. They do not identify a standard catalog of graphics IP, say that every chip vendor can apply for access, or disclose the commercial terms. The distinction is important: a statement that Nvidia licenses IP does not, by itself, mean a competitor can license an Nvidia GPU design.
What NVLink Fusion lets custom-chip designers do
Announced by Nvidia on March 31, 2026, NVLink Fusion is intended to let customers developing custom XPUs build heterogeneous AI infrastructure compatible with Nvidia systems. In practical terms, it provides an integration path for a customer’s own accelerator within a larger Nvidia-centered system. The announcement is about system compatibility and interconnect, not a general offer to license Nvidia graphics cores.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →#1 Best Overall
The design can let an organization differentiate its accelerator while using Nvidia technology for other parts of the infrastructure. Depending on the system, those surrounding components may include Nvidia CPUs, GPUs, networking, storage connectivity, software or complete systems. NVLink Fusion therefore gives Nvidia a way to participate in infrastructure built around a chip it did not design itself.
How the Nvidia partnerships differ
The public examples do not all describe the same kind of relationship. Marvell is specifically associated with NVLink Fusion and custom-XPU and networking integration. Nvidia and MediaTek, by contrast, announced a deeper collaboration on August 31, 2026, spanning AI infrastructure, local AI computing and automotive; their announcement describes joint work involving Nvidia accelerated computing, AI, graphics and software. It does not say MediaTek received a general license to Nvidia graphics cores.
Rank #2
- Ideal for Gifting
- Ideal for a bookworm
- Compact for travelling
| Company or example | Technology or work described | Partner role and integration point | What the public description establishes |
|---|---|---|---|
| Marvell | NVLink Fusion; custom-XPU and networking integration | Custom-XPU and networking integration within Nvidia-compatible infrastructure | Explicitly tied to NVLink Fusion; public commercial terms are not stated. |
| MediaTek | Collaboration involving AI infrastructure, local AI computing, automotive, accelerated computing, AI, graphics and software | Multi-generation collaboration across edge, automotive and AI platforms | A broad collaboration was announced on August 31, 2026; a general Nvidia graphics-IP license is not stated. |
Why Nvidia might enable rival silicon
A custom accelerator does not have to replace every part of an AI system. If it connects into infrastructure that uses Nvidia components, Nvidia can remain involved even when another company supplies the XPU. That can expand the range of systems built around Nvidia’s platform and preserve demand for related components or software.
The Register reported on August 31, 2026, that Nvidia was inviting competitors to license or use parts of its IP holdings, particularly networking technology. It quoted Dion Harris, Nvidia’s senior director of HPC and AI Hyperscale Infrastructure, describing possible adoption across the CPU, GPU, scale-up fabric, Spectrum-X and InfiniBand networking, and DPU capabilities. This points to a platform strategy broader than graphics alone: Nvidia may offer ways to use or integrate selected technologies across several layers of a system.
Free tools Windows power users keep installed
One-click scans. No signup required.
For Nvidia, the trade-off is that a customer can develop more of its own silicon while still building around Nvidia infrastructure. For a chip partner, the potential benefit is compatibility with that infrastructure; the corresponding trade-off is continued reliance on Nvidia components or technology at other system layers. The public announcements do not quantify either effect.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What this means for Nvidia’s business—and what remains unknown
From a business perspective, licensing or exposing selected IP could create value through partnerships and keep Nvidia technology in systems that include third-party accelerators. But the available disclosures do not establish how much revenue licensing generates, how many licensees there are, or whether any particular deal is profitable. They also do not publish royalty rates, license fees, exclusivity, minimum commitments, eligibility rules or a public application portal.
Rank #4
- Established: Nvidia uses license and development agreements, and NVLink Fusion is presented as an integration route for custom XPUs into Nvidia-compatible infrastructure.
- Not established: a blanket program for licensing Nvidia graphics cores to competing chip vendors, or a public catalog of available IP.
- Not disclosed: the financial terms, access criteria and scale of any licensing initiative.
Investors and customers should therefore distinguish a technology partnership or system-compatibility announcement from evidence of a broad, standardized GPU-IP licensing business. Nvidia has described ways for outside silicon to connect to its platform; the public material does not show that it has opened its graphics designs generally to competitors.
Quick Recap
Best Value
- It can be a gift option
- Comes with secure packaging
- Helpful in various ways
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
