Dec 3, 2024, 12:00 AM
Dec 3, 2024, 12:00 AM

Apple relies on Amazon AI chips for its search services

Highlights
  • Apple has utilized Amazon Web Services' custom AI chips for services including search, improving operational efficiency by 40%.
  • At the AWS Reinvent conference, Apple indicated that it plans to evaluate Amazon's new Trainium2 chip for pretraining its AI models.
  • This collaboration marks a significant step for Apple towards enhancing its AI capabilities, potentially affecting AI spending dynamics in the cloud services market.
Story

In the United States, Apple recently announced at the AWS Reinvent conference that it has been using Amazon Web Services' (AWS) custom artificial intelligence chips, specifically the Inferentia and Graviton chips, for various services such as search through its applications like Siri and Apple Music. Benoit Dupin, who is Apple's senior director of machine learning and AI, detailed how the use of these chips had resulted in a substantial 40% increase in efficiency for their operations. The strong collaboration between Apple and AWS over the past decade highlights Apple's growing reliance on cloud services to improve its AI capabilities, which could put pressure on competitors like Microsoft Azure and Google Cloud. Furthermore, during the conference, Dupin indicated that Apple plans to explore utilizing Amazon's latest Trainium2 chip for pretraining proprietary AI models, particularly for its Apple Intelligence service. According to Dupin, initial evaluations suggest that the Trainium2 chip might offer up to 50% improvement in efficiency during the pretraining phase of AI model development. This finding underscores that Amazon's AI chips not only provide a cost-effective solution for inference, compared to traditional x86 processors from Intel and AMD, but they also present viable options for developing and training new AI systems. The announcement about the use of these chips comes at a time when the tech industry is fiercely competing to create efficient alternatives to the dominant Nvidia graphics processors, which are currently the most popular option for AI training tasks. This move by Apple may signal other companies that there are viable alternatives to Nvidia's solutions, which can encourage investments in similar technologies and diversify AI processing options across the industry. AWS's CEO, Matt Garman, noted that Apple had been one of the initial adopters and beta testers for the Trainium chips, highlighting the importance of Apple's input to AWS in developing its Generative AI infrastructure. Apple's recent venture into leveraging custom AI chips from Amazon signifies a shift in the competitive landscape of cloud services for AI, as companies look to provide more efficient and cost-effective solutions in response to the growing demand for AI capabilities. With Appleā€™s commitment to advancing its AI initiatives and collaborating with AWS, it indicates a long-term investment in innovative technologies that can support next-generation AI models. As the tech industry moves forward, companies that harness the power of efficient cloud AI chips could potentially lead the way in AI development and deployment.

Opinions

You've reached the end