Computer processor chip representing AI infrastructure

Nvidia Puts $3.5 Billion Behind MediaTek’s Custom AI Chips

Nvidia is investing $3.5 billion in Taiwanese chipmaker MediaTek and giving it access to NVLink Fusion, a move that lets cloud providers and AI labs build custom silicon while keeping it attached to Nvidia’s data-centre architecture. TechCrunch reports that MediaTek will use Nvidia technology to design chips that plug directly into Nvidia-based systems. The deal addresses a strategic threat to Nvidia: Amazon, Google, Microsoft, OpenAI and Anthropic are all investing in chips tailored to their

Nvidia is investing $3.5 billion in Taiwanese chipmaker MediaTek and giving it access to NVLink Fusion, a move that lets cloud providers and AI labs build custom silicon while keeping it attached to Nvidia’s data-centre architecture. TechCrunch reports that MediaTek will use Nvidia technology to design chips that plug directly into Nvidia-based systems.

The deal addresses a strategic threat to Nvidia: Amazon, Google, Microsoft, OpenAI and Anthropic are all investing in chips tailored to their own workloads. Instead of trying to defeat every custom accelerator, Nvidia is making its networking and rack-scale platform the common layer around them. MediaTek said in June that its custom data-centre ASIC business could generate $2 billion in 2026 revenue, giving the partnership a sizeable commercial base. The arrangement also reflects how AI infrastructure is being bought: customers want chips tuned for particular workloads, but still need high-speed links, software compatibility and rack designs that can be deployed repeatedly at scale. Nvidia is betting those integration costs will keep its platform central even as the processor mix becomes less uniform.

Unique insight

Nvidia’s answer to custom silicon is to make custom silicon dependent on Nvidia’s interconnect and data-centre operating model.

Analysis

That changes the chip contest from a simple GPU-versus-ASIC fight into a platform contest. Hyperscalers gain more freedom to optimise cost and performance, while Nvidia preserves demand for the surrounding networking, software and rack infrastructure that makes those chips useful at scale. The near-term signal is customer adoption of NVLink Fusion; the longer-term risk for Nvidia is that partners eventually use the platform to build enough independence to weaken its pricing power.

Bottom line — If I only remember one thing tomorrow, what should it be?

Nvidia is defending its AI moat by owning the connection between chips, even when those chips are designed by someone else.

  • Story type: Breaking news
  • Impact: High
  • Confidence: Medium (multiple reports)
  • Audience: Business
  • What changed: Nvidia committed $3.5 billion to MediaTek and extended NVLink Fusion to its custom-AI-chip operation.
  • What to watch: Which hyperscalers or model companies adopt MediaTek-built silicon on Nvidia infrastructure.
  • Reading time: 2 minutes