Nvidia has announced its next-generation Rubin AI accelerator chips are nearing launch, marking a significant leap in computing power for artificial intelligence. CEO Jensen Huang revealed the update during a keynote at the Consumer Electronics Show (CES) in Las Vegas, stating that all six Rubin chips have returned from manufacturing partners and are on schedule for customer testing later this year.
The Rubin platform, successor to the Blackwell generation, promises a 3.5x improvement in AI model training speed and a 5x boost in running AI software. It features a new central processing unit (CPU) with 88 cores, offering twice the performance of its predecessor, making it better suited for complex AI workloads like planning and reasoning. Huang emphasized the "intense race" in AI, noting that "the amount of computation necessary for AI is skyrocketing" and driving unprecedented demand for Nvidia's GPUs.
The increased efficiency is also expected to lower operational costs. Nvidia highlighted that Rubin-based systems will be cheaper to run than Blackwell systems due to requiring fewer components, saving data centers on energy. Major cloud providers like Microsoft, Google Cloud, and Amazon Web Services (AWS) are anticipated to be among the first to deploy Rubin hardware in the second half of 2026.
The news has implications for the cryptocurrency sector, particularly Bitcoin mining. The article notes that the surging demand for AI computing power could drive more Bitcoin mining companies to pivot partially or fully to AI operations. This trend has been growing as Bitcoin mining difficulty increases, and AI presents an opportunity for miners to maximize their existing computational resources and seek alternative revenue streams.
Huang's presentation at CES also underscored the escalating competition in the chip market, with rivals like AMD also making significant announcements. However, Nvidia remains confident in its leading position, banking on the long-term AI market which it believes could be worth trillions of dollars.
