Apple is planning to purchase over 20,000 servers for AI through 2024, according to a new report from TF International Securities analyst Ming-Chi Kuo.
My latest survey indicates that Apple plans to purchase 2,000–3,000 and 18,000–20,000 units of AI servers in 2023 and 2024, respectively. It represents about 1.3% and 5% of worldwide AI server shipments in 2023 and 2024, respectively.
Kuo believes for 2023, Apple will be buying servers with Nvidia's HGX H100 8-GPU. These are designed for generative AI training and inference. By 4Q24, the company will likely switch to the B100. These servers are incredibly expensive, around $250,000 each!
Each HGX H100 8-GPU server is priced at around $250,000. Therefore, it's estimated that Apple will spend at least about $620 million in 2023 and $4.75 billion in 2024 on AI server purchases.
Despite these plans, Apple is still well behind the competition. Meta is expected to purchase around 40,000 servers in 2024 and its current count far exceeds Apple's.
Kuo notes that Apple prefers to use its own servers for training large language models, rather than relying on virtual hosts. He expects that Apple will have to invest at least several billion dollars annually to catch up with its competitors.
If Apple really intends to spend only a billion dollars a year on generative AI development, it won't matter much if my survey is wrong, but I am genuinely concerned about the future of Apple's generative AI business/service.
Finally, Kuo says he has not seen evidence that Apple is developing its own AI server chip which could help it save on AI server procurement costs.
Please download the iClarified app or follow iClarified on Twitter, Facebook, YouTube, and RSS for more AI updates.
My latest survey indicates that Apple plans to purchase 2,000–3,000 and 18,000–20,000 units of AI servers in 2023 and 2024, respectively. It represents about 1.3% and 5% of worldwide AI server shipments in 2023 and 2024, respectively.
Kuo believes for 2023, Apple will be buying servers with Nvidia's HGX H100 8-GPU. These are designed for generative AI training and inference. By 4Q24, the company will likely switch to the B100. These servers are incredibly expensive, around $250,000 each!
Each HGX H100 8-GPU server is priced at around $250,000. Therefore, it's estimated that Apple will spend at least about $620 million in 2023 and $4.75 billion in 2024 on AI server purchases.
Despite these plans, Apple is still well behind the competition. Meta is expected to purchase around 40,000 servers in 2024 and its current count far exceeds Apple's.
Kuo notes that Apple prefers to use its own servers for training large language models, rather than relying on virtual hosts. He expects that Apple will have to invest at least several billion dollars annually to catch up with its competitors.
If Apple really intends to spend only a billion dollars a year on generative AI development, it won't matter much if my survey is wrong, but I am genuinely concerned about the future of Apple's generative AI business/service.
Finally, Kuo says he has not seen evidence that Apple is developing its own AI server chip which could help it save on AI server procurement costs.
Please download the iClarified app or follow iClarified on Twitter, Facebook, YouTube, and RSS for more AI updates.