Apple Weighs Return to Server Market With Custom M8 Ultra Chips for Enterprise AI

Apple is considering a significant push back into the server business after a 15-year absence, designing enterprise hardware tailored to companies and governments seeking to run AI models on their own premises rather than public cloud infrastructure. According to reporting from The Information, the company is developing servers built around its forthcoming M8 Ultra chips—with two configurations under consideration featuring either two or four chips clustered together—and has explored using Nvidia's NVLink Fusion technology to enable high-bandwidth communication between processors. A market launch is not anticipated before 2029, though the project remains subject to cancellation or significant changes.