Nvidia is putting $3.5 billion into MediaTek. This is not a typical venture investment – it's a strategic positioning move. The US company is giving the Taiwanese chip designer access to its NVLink Fusion technology, enabling MediaTek to develop custom chips for cloud providers and AI firms that plug directly into Nvidia-based data centers.
The backdrop: Amazon, Google, Microsoft, OpenAI, and Anthropic are already building their own chips to reduce dependence on Nvidia's expensive GPUs. Nvidia could view this as a threat – but doesn't. Instead, the company is turning necessity into strategy: whoever builds custom chips should build them on Nvidia's infrastructure standard. This way, Nvidia maintains control of the ecosystem, even if it no longer supplies the only chips.
The essentials
- $3.5 billion investment in MediaTek
- MediaTek gains access to NVLink Fusion, Nvidia's technology for fast chip-to-chip communication – regardless of whether chips come from Nvidia or not
- MediaTek plans to scale its custom chip business to $2 billion in revenue by 2026
- The partnership also covers vehicle AI platforms and consumer AI PCs (RTX Spark)
The ecosystem play
Nvidia is positioning itself not as a pure chip vendor, but as an infrastructure platform. Dion Harris, Nvidia's Senior Director, puts it this way:
"Nvidia is an AI infrastructure company. We expanded beyond pure computing chips years ago."
This means: whether you deploy Nvidia's H100 GPUs, Intel chips, or MediaTek custom silicon – if you standardize on Nvidia's rack-scale architecture, you're in the Nvidia ecosystem. MediaTek becomes an extension of Nvidia's reach, not its competitor.
The parallel is clear: last week, Nvidia announced a similar partnership with Amazon Web Services – without direct investment, but with the same technological integration. AWS will deploy 2 million additional Nvidia GPUs and integrate NVLink Fusion.
Who wins, who depends
| Actor | Advantage | Dependency |
|---|---|---|
| Nvidia | Remains the center of AI infrastructure; earns from every chip running on NVLink | Must tolerate competing chips |
| MediaTek | $3.5B capital; access to proven technology; growth in data center market | Binds to Nvidia's standard; less independence |
| Cloud providers | Can build custom chips that remain standardized | Continue paying for Nvidia's infrastructure software |
MediaTek brings expertise: the company already develops chips for smartphones, smart homes, cars, and wireless communications. With Nvidia backing, it can transfer this expertise into the data center market – and Nvidia benefits from MediaTek's customer relationships.
What this means for companies worldwide
For chip designers and cloud providers, this sends a clear signal: anyone building AI infrastructure cannot escape Nvidia's standards – even with custom chips. This doesn't make Nvidia less relevant; it makes it more so. At the same time, the investment shows that Nvidia is willing to share – as long as partners integrate into its ecosystem.
Companies planning to develop their own AI chips must understand: it's not just about the chip itself, but the ability to integrate it into existing infrastructure. Nvidia's strategy makes that integration a prerequisite – and MediaTek shows how to become a partner rather than a competitor.
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
