In a surprising move, Apple (AAPL.O) has chosen to rely on Google's tensor processing units (TPUs) instead of Nvidia's renowned graphics processing units (GPUs) for building key components of its artificial intelligence (AI) software infrastructure. This decision, revealed in an Apple research paper published on Monday, highlights Apple's strategic shift in choosing cloud infrastructure for its forthcoming AI tools and features.
Key Details of Apple's AI Infrastructure
Apple's research paper did not explicitly mention Nvidia, the industry leader in AI processors. Instead, the iPhone maker described using two variants of Google's TPUs: TPUv5p and TPUv4. Specifically, Apple utilized 2,048 TPUv5p chips for developing AI models that will run on iPhones and other devices, and deployed 8,192 TPUv4 processors for server-side AI models. These TPUs were organized in large clusters, optimizing the training process for Apple's AI models.
Why Google's TPUs Over Nvidia's GPUs?
Nvidia (NVDA.O) is well-known for its GPUs, which dominate the AI hardware market with an estimated 80% share. Despite Nvidia's stronghold, Apple's choice to use Google's hardware may stem from the specific benefits offered by TPUs, especially in terms of efficiency and integration within Google's cloud ecosystem. Unlike Nvidia, which primarily sells its hardware as standalone products, Google offers TPUs through its Google Cloud Platform (GCP), enabling customers to access these chips via cloud services.
This partnership allows Apple to leverage Google's robust cloud infrastructure, making it easier to deploy and scale AI models. Apple's research paper also suggested the potential for even larger and more sophisticated AI models using Google's hardware, indicating a deepening of this collaboration.
Implications and Industry Reactions
The decision to sideline Nvidia in favor of Google's TPUs marks a significant shift in the AI landscape. Apple's reliance on Google's hardware for its Apple Intelligence suite and other AI initiatives underscores a growing trend where major tech companies are diversifying their hardware dependencies, potentially shaking up the market dynamics.
While Google did not comment on the collaboration, and Nvidia declined to comment, the market reacted subtly, with Apple's stock dipping slightly by 0.1% to $218.24 on Monday. This minor fluctuation reflects a cautious investor response, likely due to the anticipation of further developments in Apple's AI strategy.
Looking Ahead
Apple's engineers highlighted the capability of Google's TPUs to support even more advanced AI models, hinting at future innovations. At the June developer conference, Apple introduced several new AI features, including the integration of OpenAI's ChatGPT technology into its software, showcasing its commitment to enhancing user experience through AI.
As Apple continues to roll out portions of Apple Intelligence to beta users, the tech community and market analysts will closely watch how this partnership with Google evolves and impacts the broader AI hardware and software ecosystem.
This development underscores a pivotal moment in the tech industry, as major
players like Apple and Google shape the future of AI infrastructure and capabilities.
Source: Reuters
Commentaires