NVIDIA to Invest 4.8 Trillion Won in Taiwan's MediaTek, Also Takes on Custom AI Chips [Global AI Briefing]

[Financial News] NVIDIA Corporation has agreed to invest $3.5 billion, or about 4.8 trillion won, in Taiwan-based semiconductor design and manufacturing company MediaTek and strengthen cooperation.
As global AI model developers such as OpenAI and Anthropic move ahead with custom chip design, NVIDIA Corporation appears to be pursuing a strategy of bringing custom AI chips into its own system.
On Aug. 31 local time, NVIDIA Corporation officially announced that it will expand cooperation with MediaTek in AI infrastructure, personal AI computing and the automotive sector, while investing $3.5 billion in convertible bonds issued by MediaTek.

The centerpiece of the partnership is NVIDIA Corporation's NVIDIA NVLink Fusion. NVLink is a technology that connects multiple AI chips at ultra-high speed so they can operate like one massive system.
Using this technology, MediaTek will help connect custom AI chips designed by cloud providers and AI companies to NVIDIA Corporation's AI systems.
In effect, MediaTek holds the key technology needed to ensure that chips designed by major AI companies can run smoothly on NVIDIA Corporation's AI systems.
NVIDIA Corporation plans to expand the scope of AI computing by combining MediaTek's chip design capabilities with its own xPU, AI software and high-speed interconnect technologies.
Industry experts say NVIDIA Corporation is seeking to strengthen its influence in the AI hardware ecosystem by securing system technology that can accommodate a wide range of custom AI chips, rather than forcing the use of its own chips, as competitors in the AI semiconductor market continue to multiply.
In other words, it is a strategy to allow AI companies' custom chips to operate within NVIDIA Corporation's interconnect technology, systems and software ecosystem.
Recently, major AI companies including OpenAI, Anthropic, Google Experience Center and Microsoft have all moved to design their own AI-specific semiconductors in order to reduce dependence on NVIDIA Corporation's xPU and cut computing costs, fueling speculation that NVIDIA Corporation's influence in the AI semiconductor market may be weakening.

[email protected] Lee Gu-soon Reporter