New Delhi, July 20: Google is developing a new artificial intelligence chip aimed at making its Gemini models more efficient, highlighting the technology industry’s growing push to design specialised hardware for advanced AI systems.
The move reflects Google’s broader strategy of combining its AI software with custom-designed processors. Developing dedicated silicon can allow technology companies to optimise computing resources for specific workloads while reducing reliance on general purpose hardware.
AI models require enormous computing capacity for training and running responses. As demand for generative AI services continues to increase, improving the efficiency of the underlying hardware has become a major focus for companies competing in the sector.
Google’s latest chip effort is particularly significant because Gemini is being integrated across a growing range of products and services. More efficient processing could help the company handle increasingly sophisticated AI workloads while managing the substantial infrastructure requirements associated with them.
The development also comes amid an intensifying race among major technology companies to control more of the AI computing stack. Companies are investing heavily in processors, data centres and networking technologies as they seek greater control over the hardware powering their AI platforms.
Custom chips can provide companies with greater flexibility over performance and energy consumption. They can also help optimise AI systems for particular tasks rather than relying exclusively on commercially available processors.
Google has already invested heavily in its own AI hardware ecosystem, making the latest development part of a longer-term strategy to strengthen its computing capabilities.
The chip race is becoming increasingly important as AI companies move from simple chatbot applications toward systems capable of handling complex reasoning, coding, multimodal tasks and autonomous operations.
For Google, improving Gemini’s efficiency could help support the expansion of AI features across its products while keeping infrastructure costs under control.
The development underscores how the next phase of the AI competition will depend not only on building more capable models but also on creating the hardware infrastructure needed to run them efficiently.