This article may contain affiliate links. Learn more on our Partner Program page.
Apple wants to catch up in artificial intelligence and is apparently investing heavily in new hardware. Despite public statements about using its own Apple Silicon technology for AI servers, new information reveals that Apple is apparently ordering Nvidia servers on a large scale. The amount is considerable, the timing surprising, and the impact not entirely without controversy.
Apple has emphasized for years the importance of data protection and control over its own infrastructure. With Apple Silicon, it created its own chip line, differentiating itself from Intel and other manufacturers. Apple wanted to focus on its own solutions, especially for AI services. Now it's emerged: The company is also using Nvidia hardware—and not just a little. This raises questions, especially regarding Apple's position on data security and its own technology.
Apple officially opted for Apple Silicon
In April 2024, it was initially reported that Apple would use its own chips – Apple Silicon – for AI servers. By June, it was assumed that entire data centers would be powered by them. In September, Craig Federighi, Apple's chief software officer, officially confirmed this strategy. Apple Intelligence, Apple's new AI system, runs on Apple Silicon. This infrastructure is crucial for ensuring user data remains protected. Only with its own hardware and a specially developed operating system can it be ensured that requests are processed securely. Federighi explained that server requests are only permitted if the signature of the software on the server has been published in a transparency protocol. This is a central component of Apple's trust model.
Nevertheless: Apple orders Nvidia servers
Despite these clear statements, new information now suggests otherwise. According to analyst Ananda Baruah of Loop Capital, Apple is currently investing around $1 billion in Nvidia systems. Specifically, this involves the purchase of approximately 250 Nvidia NVL72 servers. The price per server is between $3.7 million and $4 million. The NVL72 servers are specifically designed for generative AI and so-called large language models. According to Nvidia, each of these servers contains:
- 36 Grace CPUs
- 72 Blackwell GPUs
According to Nvidia, these servers are not yet officially available as of March 18, 2025. Apple may have pre-ordered them. This suggests that the systems will not be deployed immediately, but in the medium term.
What is the purpose of these servers?
It is unclear whether Apple is using the Nvidia servers solely for development purposes or also for operating public services. If the latter is the case, it contradicts previous statements that Apple Silicon is used exclusively. It is possible that this is a separate AI cluster that runs independently of the publicly accessible Apple Intelligence system. In his note to investors, analyst Baruah states that Apple has now officially entered the business of large AI server clusters. The partners apparently include Super Micro Computer (SMCI) and Dell – both experienced providers in the server business. Baruah suspects this could be a cluster for generative AI and large language models (LLMs).
A change of course with a signal effect
If this information is correct, Apple would be pursuing two strategies simultaneously: On the one hand, Apple Silicon will continue to be used for data protection and control in publicly used AI services. On the other hand, Apple may be building a powerful development infrastructure for new AI models with Nvidia servers. This doesn't quite fit with Federighi's statement that Apple wants to set a new standard for cloud processing with its own server architecture. And it raises questions as to why Apple relies on external hardware for highly specialized AI tasks, even though it has developed a strong in-house solution (via Investors).
What's behind Apple's billion-dollar spending on AI
Apple is apparently investing around $1 billion in Nvidia servers, despite publicly communicating a clear focus on Apple Silicon. Whether this is development infrastructure or whether these systems will also be used for public services is unclear. The only thing that is certain is that Apple wants to play a role in generative AI – and is relying on proven external technology to do so. For you as a user, this is an exciting insight into Apple's actual AI strategy, which is multifaceted and less clear-cut than the official statements suggest. (Photo by Unsplash+ / Getty Images)
