Google Collaborates with MediaTek for Next-Gen AI Chips
Google is reportedly preparing to partner with Taiwan's MediaTek to produce its next generation of Tensor Processing Units (TPUs), its own AI chips. In a report by The Information, the new collaboration is said to yield chips that will be out next year. Despite the shift, Google has not cut ties with its long-time partner Broadcom, the chip designer that has exclusively collaborated with Google on AI chips for several years.
Enhancing Competitive Advantage in AI
Google's move to partner with MediaTek is a strategic move to contain its reliance on industry leader Nvidia, which has dominated the AI chip market. Like Nvidia, Google designs its own AI server chips for internal research and development and its cloud customers. This competitive strategy allows Google to minimize its overdependence on Nvidia, which companies like Microsoft-backed OpenAI and Meta Platforms have used extensively.
MediaTek's Place in Google's AI Future
The choice to onboard MediaTek comes down to various factors, including the Taiwanese company's good relationship with TSMC, a large semiconductor producer, the Information reported. The lower cost per chip offered by MediaTek is another motivating factor that makes it an ideal partner for Google instead of Broadcom. MediaTek's onboarding is expected to enable Google to produce its AI chips affordably while ensuring it retains a competitive advantage in the fast-paced industry.
Also read: Google Expands AI in the UK with New Data Residency and Startup Support
Google's Increasing Investment in AI Chip Development
Its massive financial outlay in TPUs highlights Google's efforts to build its AI chips. Last year, Google spent $6 billion to $9 billion on building these chips, research from Omdia has found, based on Broadcom's AI semiconductor revenue target. Google's recent efforts to expand its partnerships further consolidate its hold in the AI hardware market to offer alternatives to Nvidia's industry-leading chips.
Last year, Google released its sixth-generation TPUs to provide the firm and its cloud clients with a choice from Nvidia's chips.
