Google diversifies AI chip production
According to an exclusive report by The Information, Alphabet (GOOG, GOOGL) is preparing for a strategic partnership with Taiwanese company MediaTek (2454.TW) to develop the next generation of its AI chips — Tensor Processing Units (TPU), with production planned for next year.
This move marks a significant change in Google’s AI processor production strategy, although the company is not severing ties with Broadcom (AVGO) — its exclusive AI chip development partner in recent years. The information was confirmed by an employee of the San Jose-based company Broadcom.
Google’s decision to develop its own server AI chips, like Nvidia (NVDA), gives the company a significant competitive advantage in the artificial intelligence race. This approach allows it to reduce dependence on Nvidia at a time when competitors, including Microsoft (MSFT)-backed OpenAI and Meta Platforms (META), are facing growing demand for Nvidia chips.
Late last year, Google introduced the sixth generation of TPU, aiming to provide itself and Google Cloud customers with an alternative to Nvidia processors, which remain the most in-demand in the industry. The upcoming partnership with MediaTek could significantly strengthen Google’s position in the AI chip market.
Representatives from Google, MediaTek, and Broadcom have not yet provided comments regarding the potential collaboration. However, this development reflects a growing trend toward supply chain diversification in the technology sector, especially in the critical area of AI computing.
