At AWS Reinvent 2024, Apple publicly revealed for the first time the details of its partnership with Amazon Web Services (AWS) in the field of artificial intelligence. This rare collaboration shows that Apple is adopting Amazon's custom AI chips, such as Inferentia, Graviton, and the latest Trainium2, to improve the efficiency of its machine learning and AI services, especially in key apps such as search, voice assistant Siri, Apple Maps, and Apple Music.
The background of the cooperation and Apple's AI needs
Apple has long used Amazon AWS's cloud infrastructure to support the global deployment of its services. For example, products such as Apple's Siri voice assistant, Apple Maps, and Apple Music all rely on cloud computing resources provided by AWS. Benoit Dupin, Apple's senior director of machine learning and artificial intelligence, mentioned during the conference that AWS's infrastructure plays a vital role in service reliability and global customer support, and its custom chips improve the efficiency of AI inference.
Apple's partnership with Amazon has focused on AWS's AI-specific chips, Inferentia and Graviton, which are already being used in Apple's search service and are making them 40 percent more efficient. This achievement reflects the important role of customized hardware for specific AI application scenarios.
Apple's collaboration with Amazon's Trainium2 chip
With the development of AI technology, Apple not only uses Amazon's chips in the inference process, but also begins to evaluate the Trainium2 chip for more efficient AI model pre-training. Compared to Intel's x86 processors and AMD's processors, Amazon's Trainium2 chip offers significant advantages in computing power and cost-effectiveness, especially in large-scale machine learning tasks.
Pictured: Apple in partnership with Amazon (Source: CnBta).
According to Dupin, Apple expects to improve efficiency by 50 percent by using the chips in the pre-training of the model in the early stages of evaluating the Trainium2 chips. This data further validates Amazon's technical strength in the field of AI hardware and Apple's interest in custom hardware. Apple's use of the Trainium2 chip this time also marks its further innovation and deepening in artificial intelligence technology.
Apple's vision for generative AI
In recent years, generative AI has become a hot topic in the tech industry. Apple has also made active explorations in this area. This fall, Apple released its first generative AI product, Apple Intelligence, a suite of AI services that can aggregate notifications, rewrite emails, generate new emojis, and more. Apple plans to integrate with OpenAI's ChatGPT in the future, and has predicted that Siri will get stronger natural language processing and app control capabilities next year.
Unlike traditional chatbots, which rely on large clusters of Nvidia GPUs for training, Apple has taken a different path: whenever possible, it leverages M-series chips on local devices, such as iPhones, iPads, and Macs, to handle AI tasks. Complex queries are sent to Apple's own servers for more in-depth processing. This approach not only optimizes processing efficiency, but also ensures stronger privacy protection and device localization.
The dual advantages of cloud services and custom chips
Apple's partnership with Amazon demonstrates the great potential of combining cloud computing with custom hardware. By using the efficient custom AI chips provided by AWS, Apple is able to improve the efficiency of AI computing while ensuring the scalability of cloud services. This collaborative relationship is also an important achievement for AWS and a testament to Amazon's ability to innovate in the AI hardware space.
As more and more companies seek to reduce costs in the AI field, Amazon's Trainium2 chip and Graviton processors provide powerful computing power support for AI applications. Apple's addition could inspire more companies to turn to non-Nvidia AI hardware solutions, creating a new competitive landscape in the market.
A vision for the future
Apple's partnership with Amazon marks an increasingly complex collaboration and competition between tech giants in the field of AI hardware. With the rise of generative AI, more and more companies need to rely on efficient and flexible hardware support to handle increasingly complex AI model training and inference tasks. Apple's cooperation with Amazon shows its determination to continue to increase AI technology, and it also marks a new stage in Apple's AI development.
In the future, as generative AI technology continues to advance, Apple Intelligence is expected to bring more innovative features to Apple, improve the user experience, and compete fiercely with competitors such as Microsoft Azure and Google in the field of AI and cloud computing.
Conclusion
Apple's strategic layout in the field of AI chips has been further clarified. By partnering with AWS and adopting chips like its custom Trainium2, Apple is not only optimizing the efficiency of its AI applications, but also demonstrating its ambitions in the field of generative AI. As Apple continues to advance its AI technology, especially in generative AI and natural language processing, its competitiveness in the fields of AI hardware, cloud computing, and smart devices will undoubtedly be greatly improved in the future. For the tech industry as a whole, this cross-platform, cross-company collaboration model also provides a fresh perspective on future development.
