Nvidia is deepening its alliance with Taiwanese chip designer MediaTek with a $3.5 billion investment in convertible bonds, the two companies announced Monday, August 31, 2026. Alongside the capital commitment, MediaTek is adopting Nvidia's NVLink Fusion platform — a move that positions the world's largest smartphone chip designer as a key manufacturing partner for the hyperscalers and AI labs now building their own custom silicon.
The announcement, made in a joint press release, extends the companies' long-standing collaboration across three areas: cloud AI infrastructure, local AI computing and automotive platforms. For readers tracking AI industry coverage, the deal is a clear signal of where Nvidia sees the next wave of growth — and how it plans to keep custom chips tethered to its ecosystem rather than competing against it.
Why the Custom Chip Market Needs NVLink
The biggest tech companies — hyperscalers, cloud providers and frontier model developers — increasingly want AI accelerators designed around their own workloads rather than off-the-shelf GPUs. But designing a custom XPU is only half the battle. Deploying hundreds or thousands of them inside a data center requires high-speed interconnects, memory architecture, advanced packaging and rack-scale systems engineering that most chip buyers cannot build alone.
That is the gap NVLink Fusion is designed to fill. According to the press release, MediaTek will offer the NVLink Fusion platform as a design foundation for customers developing custom AI accelerators, giving them a "prebuilt, prequalified and system-prevalidated foundation for multi-die XPU development" and a path into Nvidia's NVLink-connected, rack-scale AI factories.
In practical terms, customers can bring their XPU designs to MediaTek and tailor connectivity, memory, packaging, performance and power characteristics to their workloads, while relying on the two companies for the NVLink connectivity, memory architecture, packaging, manufacturing and rack-scale technologies required for actual deployment. Nvidia says the approach lets customers focus resources on the differentiated compute that defines their platforms instead of re-engineering every surrounding system.
MediaTek's designs would also plug into Nvidia's MGX rack-scale architecture, which the companies say reduces development complexity, increases performance and shortens time to market for semi-custom AI factories.
What Nvidia's Money Buys
The $3.5 billion takes the form of convertible bonds issued by MediaTek — a structure that makes Nvidia a creditor today and a potential shareholder later, deepening the integration between the two companies without an outright acquisition. HotHardware, which reviewed the announcement, described it as a way for Nvidia to "deeply integrate MediaTek into its hardware roadmap" while keeping its own ecosystem at the center of the custom AI chip push.
MediaTek brings serious credentials to the arrangement. The press release cites the company's leadership in custom silicon, high-performance computing, power-efficient system-on-chip design, advanced packaging, interconnects and connectivity — the exact skill set required to turn a custom accelerator concept into a manufacturable product.
"AI is transforming every computing platform — from the world's largest AI factories to the PC and the car," said Jensen Huang, Nvidia's founder and CEO, in the announcement. "MediaTek is one of the world's great semiconductor companies, with exceptional expertise in system-on-chip design, connectivity, leading performance and power efficiency. Together, we're building platforms that bring NVIDIA accelerated computing to new markets."
MediaTek vice chairman and CEO Rick Tsai said the investment strengthens a collaboration spanning "cloud AI infrastructure, local AI computing and automotive in the era of physical AI," combining Nvidia's accelerated computing leadership with MediaTek's edge-to-cloud portfolio.
Beyond the Data Center: PCs and Cars
The partnership stretches well past rack-scale infrastructure. In local AI computing, Nvidia and MediaTek are collaborating on platforms for generative and agentic AI running on devices rather than in the cloud — an area HotHardware reports includes future generations of Nvidia's RTX Spark and DGX Spark systems, which pair Nvidia silicon with MediaTek's power-efficient SoC design expertise.
Automotive is the third pillar. The plan is to combine MediaTek's automotive system-on-chip design experience with Nvidia's DRIVE AGX hardware and software stack, targeting the growing compute demands of software-defined vehicles and physical AI.
MediaTek is not the only design partner Nvidia is courting. The company confirmed that MediaTek joins a broader NVLink Fusion ecosystem that already includes SiFive, which is likewise adopting the platform — part of a deliberate strategy to make NVLink the default interconnect for whatever custom chips emerge from the current wave of in-house silicon projects at cloud giants and AI labs.
What It Means for the AI Hardware Market
The custom XPU race is often framed as a threat to Nvidia: if Google, Amazon, Meta, Microsoft, OpenAI and others build their own accelerators, why would they keep buying GPUs? Nvidia's answer, embodied by NVLink Fusion and now reinforced by a $3.5 billion stake in MediaTek, is to make sure that even Nvidia's competitors' chips depend on Nvidia's networking, packaging and rack-scale technology.
Every custom XPU that connects through NVLink still generates revenue for Nvidia — and keeps alternative interconnect standards on the sidelines. With MediaTek's manufacturing reach spanning smartphones, PCs, vehicles and data centers, Nvidia has effectively hired one of the world's most prolific chip designers as the gateway for custom silicon into its ecosystem.
The deal also lands at a moment of extraordinary momentum for Nvidia, which recently reported $96.2 billion in quarterly revenue, up 106% year over year. Betting on MediaTek is a bet that the custom chip boom widens the market — and that Nvidia intends to profit from it either way.
Stay Ahead of AI Read more AI news →
