Nvidia Rubin chips are set for mass production this year, and computing power costs are expected to decrease significantly.

【BlockBeats】NVIDIA has announced a major move: the highly anticipated Rubin data center chips are set to officially launch this year. These six chips have been delivered from manufacturing partners and have passed multiple critical tests, indicating that product deployment is proceeding smoothly.

In terms of performance comparison, the new generation Rubin accelerators have shown significant improvements over the previous Blackwell chips—training performance has increased by 3.5 times, and AI software running speed has improved by 5 times. But the most exciting aspect is the cost advantage: Rubin systems can achieve the same results with fewer components, which means that systems based on Rubin will have significantly lower operating costs compared to Blackwell solutions.

According to sources, Microsoft and other major cloud service providers have already secured quotas in advance, planning to deploy this new hardware in the second half of the year. For the entire AI infrastructure industry, this means data center operational efficiency will be further enhanced, and the potential for reducing computing costs will also be unlocked.