OpenAI Weighs Buying Google AI Chips to Reduce Nvidia Dependence
Summary
- OpenAI is considering adopting Google AI TPU chips as part of a strategy to expand its AI infrastructure and reduce its reliance on Nvidia.
- OpenAI has already said it plans to begin using its in-house AI chip, Jalapeno, which it developed with Broadcom, later this year.
- If OpenAI completes the purchase of Google chips, the Nvidia-centered structure of the AI chip market could ease as supply-chain diversification expands.
Forecast Trend Report by Period
OpenAI is considering buying Google-developed chips as it expands its artificial intelligence infrastructure.
Bloomberg reported on October 27 that OpenAI is reviewing the adoption of Tensor Processing Units, or TPUs, Google's AI chips. Specific contract terms have yet to be finalized, and neither company has confirmed the discussions.
The move is part of a strategy to secure the computing power needed to develop and run AI models while reducing reliance on Nvidia chips. OpenAI has already said it plans to begin using its own AI chip, Jalapeno, developed jointly with Broadcom, later this year.
Richard Ho, who leads OpenAI's in-house chip development team, previously worked on Google's TPU program. Google began developing TPUs more than a decade ago to improve the speed and efficiency of its search services. More recently, it has expanded the business into AI computing chips that can replace Nvidia's graphics processing units, or GPUs.
The use of Google's TPUs is also widening. Oracle said on October 27 that it plans to offer Google TPUs through its cloud infrastructure. Amin Vahdat, a Google executive, emphasized that next-generation AI development requires a variety of computing resources.
If OpenAI completes the purchase of Google's chips, it would further accelerate supply-chain diversification in an AI chip market centered on Nvidia.
Suehyeon Lee
shlee@bloomingbit.ioI'm reporter Suehyeon Lee, your Web3 Moderator.