NVIDIA Corp. is advancing its next-generation AI computing platform, Rubin, with volume production expected in the second half of 2026. The Rubin platform, described as the company’s third-generation NVLink Rack Scale AI supercomputer, features six new chips that have all been taped out and sent to production at Taiwan Semiconductor Manufacturing Company (TSMC). These chips utilize TSMC's 3nm (N3P) process and employ advanced packaging technologies such as CoWoS-L and CoPoS. NVIDIA’s Chief Financial Officer confirmed Rubin remains on schedule for volume production next year, supported by a mature and fully scaled supply chain. Industry reports suggest mass production could begin ahead of schedule in the fourth quarter of 2025, with packaging potentially starting by year-end. Rubin is NVIDIA’s first product to adopt a chiplet architecture, a development that prompted CEO Jensen Huang to personally oversee progress. The company is also preparing to introduce the Vera CPUs and companion chips alongside Rubin as part of its upcoming NVL144 platform for AI data centers. Customers for these chips include SK Telecom and KT. Meanwhile, NVIDIA’s gaming division continues to post record revenues.