NVIDIA´s highly anticipated ´Rubin´ Artificial Intelligence GPU project has reportedly surged ahead of schedule, according to multiple industry sources close to the development. Initial rumors about the ´Rubin´ accelerators targeting a late 2025 launch have gained new credibility, as insiders suggest the company´s engineering teams have outpaced internal targets by nearly six months. Despite this accelerated progress, analysts do not expect NVIDIA to expedite a public release, particularly with the recent debut of ´Blackwell Ultra´ hardware, which remains in the spotlight for current deep learning and data center deployments.
Key to the ´Rubin´ roadmap is NVIDIA´s continued collaboration with Taiwan Semiconductor Manufacturing Company (TSMC), which is set to manufacture the R100 GPUs using its advanced 3 nanometer N3P process and employ CoWoS-L packaging for optimal performance and scaling. Reports from Taiwan indicate the project is entering a critical phase, with final ´taping out´ processes for both Rubin GPUs and the related Vera CPUs slated for completion this month. Trial production across the summer is expected to yield initial ´Rubin´ samples by September, a substantial milestone that positions the chips well for a public reveal in early 2026.
Supply chain sources cited in a recent Ctee TW report highlight that the entire ´Rubin´ and Vera chip development process has proceeded more smoothly than previous NVIDIA efforts. Mass production for both is scheduled to start in 2026, aligning with predictions that finished products may be showcased for the first time at high-profile industry events such as CES 2026. Collectively, these developments underline NVIDIA´s consistent momentum in the Artificial Intelligence hardware space, as the company prepares for the next era beyond its present ´Blackwell´ family of processors.
