Google is reportedly preparing to collaborate with Taiwan’s MediaTek on the next generation of its AI chips—its Tensor Processing Units—set to be produced next year, according to a report by Information citing people involved in the project.

The move is seen as part of Google’s ongoing efforts to boost its AI capabilities and reduce its reliance on competitors like Nvidia. Unlike earlier projects where Google worked exclusively with Broadcom on its AI chip designs, the tech giant is now considering MediaTek, which offers competitive pricing and maintains a strong relationship with TSMC. A San Jose-based employee mentioned that despite the new partnership, Google has not cut ties with Broadcom.

Google designs its own AI server chips for internal research and development and also rents them out to its cloud customers. This strategy is intended to give the company a competitive edge, particularly as rivals such as Microsoft-backed OpenAI and Meta Platforms experience soaring demand for Nvidia chips. In line with these efforts, Google rolled out its sixth-generation TPU late last year, positioning itself with an alternative to Nvidia’s processors.

MediaTek’s appeal lies partly in its cost-effectiveness; it charges Google less per chip compared to Broadcom. This factor, combined with MediaTek’s robust ties to TSMC, makes it an attractive partner as Google looks to scale up its AI chip production. Research firm Omdia estimated that Google spent between $6 billion and $9 billion on TPUs last year, reflecting the significant investment in its AI infrastructure.

Neither Google, MediaTek, nor Broadcom immediately responded to requests for comment on the matter. As the AI race intensifies, industry analysts say such partnerships are crucial for maintaining technological leadership in a rapidly evolving market.

Copied