Apple has publicly stated that it is using Apple Silicon servers to support the operation of Apple Intelligence, but according to Loop Capital analyst Ananda Baruah, the company is now also spending $1 billion to purchase NVIDIA's AI systems.
"AAPL has officially joined the large server cluster GenAI game, with SuperMicro [SuperMicroComputer] and Dell as key server partners," he wrote in a note to investors. "While we are still gathering more comprehensive information, this seems likely to become a GenAILLM (Large Scale Language Model) cluster."
Baruah claimed that Apple is purchasing 250 NVIDIA NVL72 servers, with each server costing between $3.7 million and $4 million.
According to NVIDIA, its NVL72 server contains 36 GraceCPUs and 72 Blackwell GPUs. The company also said that as of March 18, 2025, the server is not yet available.
There's no doubt that Apple can pre-order servers now, and it's no surprise that the company feels the need to expand its servers. Judging by the numbers, this could be for development purposes rather than for the public, but there's no way of telling right now - and that's assuming the reports are correct.
If it's intended for more than just development, that doesn't quite square with Federighi's assertion that using Apple Silicon servers "sets a new standard for cloud processing in the industry."
"Building servers in the data center and building a custom operating system that runs in the data center was a huge undertaking when we didn't have Apple Silicon servers before," he said. "[Creating] a trust model where your device will deny requests to the server unless signatures of all the software the server is running have been published to the transparent logs is certainly one of the most unique elements of the solution and critical to the trust model."