Apple Invests $1 Billion in Nvidia Servers to Enhance AI Capabilities Amid Development Challenges

Analyst Ananda Barua from Loop Capital has reported that Apple is in the process of placing orders for Nvidia’s GB300 NVL72 systems, amounting to approximately $1 billion. The company plans to acquire around 250 servers, each costing between $3.7 million and $4 million.

This report comes amid recent news regarding Apple’s challenges in developing its proprietary artificial intelligence system.

Nvidia is set to offer several ready-made systems powered by their new computing accelerators in the latter half of 2025. The GB300 NVL72 is a preconfigured server rack that integrates 72 Blackwell Ultra GPUs and 36 Grace CPUs. It features a liquid cooling system, utilizes fifth-generation NVLink, includes Nvidia’s ConnectX-8 SuperNIC modules, and boasts 18 TB of LPDDR5X RAM. Its performance reaches up to 1100 petaflops in FP4 calculations and nearly 1400 petaflops for sparse computations.

Apple is collaborating with server manufacturers Dell Technologies and Super Micro Computer to build a substantial server cluster to support generative AI applications.

Earlier reports indicated that Apple was developing its own chip intended for server use. However, it appears that progress has either slowed down more than anticipated, or the company has entirely abandoned this initiative.

Meanwhile, a class-action lawsuit has been filed against the company in the United States over delays in the release of Apple Intelligence on iPhones and other devices, despite the company promoting this AI system as a finished product. In reality, Apple has only introduced its AI in a beta version with limited functionalities and for a select group of users.

Apple presented Apple Intelligence in June 2024, but as of early 2025, it has yet to roll this feature out to stable versions of iOS.