Archive — US — techcrunch
Nvidia launches powerful new Rubin chip architecture
Nvidia launched its new Rubin chip architecture at CES 2026, designed for advanced AI computation. CEO Jensen Huang highlighted its significant improvements in speed and power efficiency, stating that it will operate up to five times faster than its predecessor, Blackwell. Rubin chips will support major cloud providers and contribute to an expected $3-4 trillion investment in AI infrastructure over the next five years.
Key points
- Nvidia officially launched the Rubin chip architecture at CES 2026.
- CEO Jensen Huang emphasized the architecture's state-of-the-art capabilities for AI hardware.
- Rubin chips are set to replace the Blackwell architecture and enhance computation capabilities.
- They are expected to produce significant speed improvements: 3.5 times faster for model training and 5 times for inference.
- Key partners include Anthropic, OpenAI, and AWS, showing broad acceptance among cloud providers.
- The new architecture consists of six chips, addressing storage and interconnection bottlenecks.
- The expected market for AI infrastructure investment is projected to reach $3-4 trillion over five years.
Why it matters
- The launch of the Rubin architecture is a response to the growing demands for AI infrastructure in a competitive market.
- Nvidia's developments echo the broader trend towards advanced AI processing technology, which is critical for future innovations.
- The architecture’s efficiency improvements align with current IT trends focusing on sustainability in tech.
- The competition for AI infrastructure mirrors past tech races, such as the early days of cloud computing.
The introduction of Nvidia's Rubin architecture marks a pivotal advancement in AI technology, addressing rising computational needs while positioning the company for significant growth as the AI infrastructure market expands.