MediaTek’s custom AI chip division is on track to generate $2 billion in revenue this year, a figure that puts the Taiwanese semiconductor company at the center of a market opportunity it estimates will reach $80 billion by 2027. The projection comes as MediaTek deepens its partnership with Nvidia through a $3.5 billion convertible bond investment announced on August 31.
The deal gives MediaTek access to Nvidia’s NVLink Fusion platform, which allows hyperscalers and other technology companies to build custom AI chips that integrate directly with Nvidia’s data center infrastructure. That access is the strategic prize: as cloud providers like Google, Amazon, and Microsoft race to design their own AI accelerators to reduce dependence on Nvidia’s off-the-shelf GPUs, they need a foundry partner that can connect their custom silicon to Nvidia’s networking and compute fabric. MediaTek wants to be that partner, and the $3.5 billion bet from Nvidia suggests the chipmaker has the capabilities to deliver on that promise.
Nvidia CEO Jensen Huang called MediaTek one of the world’s great semiconductor companies in the press release accompanying the announcement. MediaTek vice chairman and CEO Rick Tsai said the investment strengthens a collaboration that spans cloud AI infrastructure, local AI computing, and automotive in the era of physical AI. The language was notable for its breadth, suggesting the partnership goes well beyond a single product line or market segment into a multi-year strategic alignment across multiple computing categories and product lines.
The $3.5 billion investment came in the form of convertible bonds, a structure that gives Nvidia an equity stake if MediaTek’s stock reaches a predetermined price. MediaTek shares jumped 10% on the Taiwan Stock Exchange following the announcement, closing at TWD 4,315. The stock has risen from around TWD 1,500 in January to over TWD 4,300, a gain of nearly 200% for the year driven largely by optimism about the custom AI chip opportunity and the Nvidia partnership.
MediaTek is the world’s largest smartphone chipmaker by market share, according to Counterpoint Research, and the primary rival to Qualcomm in mobile processors. But the company has been diversifying aggressively into data center, automotive, and edge computing markets. Its Dimensity Auto line powers infotainment and advanced driver assistance systems in vehicles from several Chinese automakers, and the company’s RTX Spark and DGX Spark chips, developed jointly with Nvidia, target consumer PCs and enterprise workstations across multiple generations of products.
The custom AI chip market is growing faster than most analysts expected. Hyperscalers spent an estimated $180 billion on AI infrastructure in 2025, and that figure is projected to exceed $250 billion in 2026. A significant portion of that spending is flowing toward custom silicon designed for specific workloads rather than general-purpose GPUs. Google’s Tranium chips, Amazon’s Trainium processors, and Microsoft’s Maia accelerators all represent this shift, and each requires a manufacturing partner with advanced packaging and design capabilities that can deliver at scale.
MediaTek’s position in this market is unique. Unlike pure-play foundries such as TSMC, which manufacture chips designed by others, MediaTek designs and sometimes co-develops custom silicon with its customers. That model gives it closer relationships with hyperscalers and the ability to iterate quickly on design requirements and specifications. The NVLink Fusion partnership extends this advantage by ensuring that MediaTek-designed chips can plug directly into Nvidia’s compute clusters, a capability that competing foundries simply cannot offer to their customers.
The financial implications are significant. MediaTek disclosed in June that its custom AI chip division is on track for $2 billion in 2026 revenue. If the broader market reaches $80 billion by 2027 as the company projects, even a modest market share gain would translate into billions in additional revenue. The company’s total revenue was approximately $19 billion in 2025, so the custom AI chip business could represent 10% or more of the top line within two years, a meaningful shift for a company that has historically relied on smartphone SoCs for the bulk of its income.
The partnership also extends into consumer and automotive markets. Nvidia and MediaTek are collaborating on multiple generations of RTX Spark and DGX Spark chips for consumer PCs and AI developer workstations. In automotive, MediaTek’s Dimensity Auto chips will integrate with Nvidia’s accelerated computing platforms for software-defined vehicles. The breadth of the collaboration suggests that both companies see MediaTek as a long-term strategic partner rather than a transactional supplier in any single market segment or product category.
The deal also positions MediaTek competitively against Qualcomm, which has its own custom AI chip ambitions through its Cloud AI 100 accelerator. While Qualcomm has focused on edge inference, MediaTek’s NVLink Fusion access gives it a direct line into data center training clusters, a market that is both larger and faster growing. The competitive dynamic between the two companies is likely to intensify as the custom silicon market expands toward the $80 billion total addressable market that MediaTek has identified for 2027.
For the broader semiconductor industry, the MediaTek-Nvidia deal signals that custom AI silicon is moving from a niche experiment to a mainstream infrastructure layer. As hyperscalers compete on AI performance, the ability to design and deploy custom chips quickly becomes a competitive advantage. MediaTek, with its combination of design expertise, manufacturing scale, and now Nvidia’s networking technology, is positioning itself as a key enabler of that transition across cloud, edge, and automotive markets in the years ahead.
discussion