In an industry where secrecy is often the name of the game, Apple made a rare move this week by openly discussing its partnership with Amazon Web Services (AWS) during the annual AWS re:Invent conference. The revelation? Apple uses Amazon’s custom AI chips, including Inferentia and Graviton, to power some of its services, and it’s evaluating the potential of Amazon’s latest chip, Trainium2, for pretraining its own AI models.

This unexpected disclosure sheds light on Apple’s evolving AI strategy and the competitive dynamics among major tech players like Amazon, Google, and Microsoft.

Benoit Dupin, Apple’s senior director of machine learning and AI, took the stage at AWS re:Invent 2024, confirming a decade-long partnership between the two companies. He described AWS as a “reliable and globally scalable infrastructure” that supports Apple’s AI-driven services like Siri, Apple Maps, and Apple Music.

The relationship extends beyond just storage and computing power. Apple has tapped into Amazon’s custom AI chips—Inferentia for inference tasks (running pre-trained models) and Graviton for other processing needs. According to Dupin, these chips have delivered significant performance improvements, including a 40% efficiency gain in some services like search.

But the real eyebrow-raiser came when Dupin hinted at Apple’s plans to evaluate Amazon’s newest AI chip, Trainium2, for pretraining proprietary models. Pretraining is a crucial step in building AI systems, requiring immense computational resources to process massive datasets.

“In early stages of evaluating Trainium2, we expect up to a 50% improvement in efficiency with pretraining,” Dupin noted.

Apple’s embrace of Amazon’s custom silicon comes at a time when AI spending is reshaping the cloud computing landscape. Amazon, Microsoft Azure, and Google Cloud are vying to capture market share in the lucrative AI segment.

Interestingly, this isn’t Apple’s first foray into third-party cloud AI solutions. Earlier this year, a research paper revealed that Apple had used Google Cloud’s Tensor Processing Units (TPUs) for training its iPhone AI system, branded as “Apple Intelligence.”

While NVIDIAs GPUs dominate the AI market, alternatives like Amazon’s Trainium and Google’s TPUs are gaining traction for their cost and efficiency advantages. Apple’s willingness to experiment with these options could signal a shift in the industry, encouraging other companies to explore non-NVIDIA solutions for their AI workloads.

Apple has been steadily ramping up its AI capabilities. This fall, it introduced its first major generative AI product, “Apple Intelligence,” which powers features like summarizing notifications, rewriting emails, and even creating new emojis.

The company is also pushing the boundaries of on-device AI. Unlike OpenAI’s ChatGPT or Google’s Gemini, which rely heavily on cloud-based NVIDIA GPU clusters, Apple’s approach focuses on performing as much AI processing as possible on its own devices. Using the M-series chips in iPhones, iPads, and Macs, Apple handles simpler AI tasks locally, while offloading more complex queries to Apple-operated servers.

This hybrid strategy reflects Apple’s commitment to privacy and efficiency, ensuring that sensitive data stays on the user’s device whenever possible.

Apple’s partnership with AWS is more than just a tech story—it’s a signal of broader industry trends. By openly endorsing Amazon’s chips, Apple is not only validating AWS as a leader in cloud AI but also highlighting the potential of custom silicon to challenge NVIDIA’s dominance.

For Amazon, having Apple as a marquee customer is a powerful endorsement as it rolls out its Trainium2 chip to a wider market. AWS CEO Matt Garman emphasized the collaborative nature of the relationship, saying, “Apple came to us with a vision for building Apple Intelligence, and we’ve worked together to make that vision a reality.”

As AI continues to shape the future of technology, Apple’s willingness to collaborate—and experiment—suggests a pragmatic approach to innovation. Whether it’s leveraging Amazon’s chips or integrating with OpenAI’s ChatGPT, Apple is positioning itself as a major player in the AI space.

The competition for AI dominance is heating up, and Apple’s calculated moves indicate it’s in it for the long haul. As it balances cloud partnerships with its own hardware innovations, Apple might just be redefining what it means to be at the cutting edge of artificial intelligence.