Google Unveils New AI Chip for Servers, Promises 30% Increase in Efficiency
Google has recently introduced a new chip called Axion, developed in-house, designed exclusively for artificial intelligence applications. Unlike other chips used in consumer products like phones and speakers, Axion is targeted for servers and promises to be 30% more efficient than existing ARM processors commonly used in these servers, and 50% more efficient than equivalent chips from Intel.
The tech giant has already started using Axion in some internal services like YouTube and Google Earth, with plans to make it available on Google Cloud and other tools for third-party companies later this year. The move signifies a shift in the artificial intelligence race among major tech companies, with rivals like Microsoft and Amazon also developing their own specialized processors for AI tasks.
The benefits of proprietary chips for artificial intelligence are clear – they offer increased processing power while reducing energy consumption, which is crucial for network infrastructure services. Google’s focus on AI technology was evident at its recent Google Cloud platform conference in Las Vegas, where the company announced updates to its artificial intelligence tool Gemini, aimed at providing more accurate responses based on specific data sources.
The conference also saw the launch of Gemini 1.5 Pro, an advanced version of the language model designed to compete with cutting-edge models like GPT 4.0 Turbo. Despite the advancements, Google is working to address concerns about the accuracy and fairness of its language model, following incidents like providing historically inaccurate images using diverse models, such as Asian or black soldiers in Nazi Germany during World War II.
Overall, Google’s introduction of Axion and advancements in AI technology demonstrate the company’s commitment to staying at the forefront of the artificial intelligence industry, offering more efficient and reliable solutions for businesses and consumers alike.