Nvidia Invests $3.5 Billion in MediaTek to Expand AI Chip Collaboration
Nvidia backs MediaTek to keep its infrastructure at the center of custom AI chip development, even as Big Tech builds its own silicon.
Key points
- Nvidia is investing $3.5 billion in Taiwanese chipmaker MediaTek.
- MediaTek will use Nvidia's NVLink Fusion technology to design custom AI chips that slot directly into Nvidia-based data centers.
- The deal extends an ongoing pattern: Nvidia investing in partners that stay inside its ecosystem.
- MediaTek expects its custom data center chip business to reach $2 billion in revenue by 2026.
What does this investment mean for AI technology?
Nvidia is putting $3.5 billion into MediaTek to help it build custom chips for AI companies and large-scale cloud providers, chips designed to plug directly into Nvidia-run data centers. The pressure driving that decision is real: Amazon, Google, Microsoft, OpenAI, and Anthropic are all developing their own chips to reduce dependence on Nvidia's GPUs (the processors that handle heavy AI computation). Nvidia's answer isn't to fight that trend. It's to be the plumbing those chips run through.
"Nvidia is an AI infrastructure company," Dion Harris, Nvidia's senior director of HPC and AI hyperscaler infrastructure solutions, told reporters Monday. "We expanded beyond pure computing chips years ago."
Why partner with MediaTek?
MediaTek brings proven expertise in chips for smartphones, automobiles, smart homes, and wireless devices. The deal gives it access to NVLink Fusion, which we first covered on 26 August as the technology that lets any chip, not just Nvidia's own, communicate quickly inside Nvidia's infrastructure. That access lets MediaTek offer cloud companies and AI labs custom chips compatible with Nvidia's full rack-scale platform rather than requiring them to rebuild their infrastructure from scratch.
How does this fit Nvidia's broader strategy?
Nvidia announced a similar tie-up with Amazon Web Services just last week, though without direct investment. AWS committed to deploying 2 million additional Nvidia GPUs and integrating NVLink Fusion, as our earlier report detailed. The MediaTek deal follows the same logic: let partners build what they want, as long as it runs on Nvidia's platform.
The two companies will also keep collaborating on DGX Spark, Nvidia's small desktop AI computer for developers, on RTX Spark for consumer AI PCs, and on software-defined vehicle platforms using Nvidia Drive AGX, the company's in-car computing system for autonomous driving.
"AI is transforming every computing platform," Jensen Huang, Nvidia's founder and CEO, said in a statement. "Together, we're building platforms that bring NVIDIA accelerated computing to new markets."
What matters most here is that Nvidia's real bet isn't on any single chip. It's that whoever builds the custom silicon will still need Nvidia's network technology underneath it, and that's a harder position to compete away than a faster GPU.
Common questions
What are GPUs and why are they important?
GPUs (Graphics Processing Units) handle the parallel calculations AI training and inference demand, far faster than standard processors.
How does this deal benefit MediaTek?
MediaTek gains access to Nvidia's infrastructure ecosystem, helping it win custom chip contracts from cloud companies and AI labs that already run on Nvidia's platform.
Will this affect consumer technology?
Directly: the two companies are already collaborating on consumer AI PCs and automotive platforms, so some of this work reaches products beyond the data center.
