Alphabet shares climbed 3% on Monday following a report by The Information detailing the development of a new, specialized AI chip. Codenamed 'Frozen v2,' the processor is being engineered specifically to enhance the efficiency of Google’s Gemini AI models.
Energy Efficiency and Technical Benchmarks
Google engineers estimate that the Frozen v2 chip will be capable of processing six to ten times more tokens per unit of energy than the company’s latest Tensor Processing Units (TPUs). According to the report, Frozen v2 is not intended to replace general-purpose TPUs but will serve as a specialized solution within Google’s custom silicon portfolio.
Addressing the Compute Shortage
The project aims to mitigate a significant internal shortage of computing power that has caused internal friction and reportedly forced Google Cloud to turn away external customers. The scale of this deficit is highlighted by Google’s recent agreement to pay SpaceX nearly $1 billion per month to secure the necessary computational resources to meet its obligations.
Alphabet aims to have the new chip operational by 2028, signaling a long-term strategic commitment to infrastructure as it struggles to keep pace with the soaring demands of generative AI.
