Summary of Key Points
NVIDIA's Vera Rubin platform, which includes the Vera CPU, NVLink networking, and NVL72 computing units, has entered the actual delivery phase. Leading AI companies (such as OpenAI and Anthropic), technology giants (like SpaceX and Microsoft), and cloud providers (Google Cloud and Oracle Cloud) are all utilizing this platform. By optimizing performance, innovating in hardware, and leveraging supply chain advantages, Vera Rubin addresses the critical issues of power consumption, cost, and scalability in AI infrastructure. It also opens up a new market worth $200 billion for NVIDIA, positioning it as the core computing infrastructure for the era of intelligent agents.
Detailed Analysis
1. Leading Customers Show Interest: Vera Platform Becomes Highly Desired in the AI Community
The Vera CPU was delivered in June and immediately received orders from top clients like OpenAI, Anthropic, and SpaceX—companies that are at the forefront of AI research and space exploration. Their adoption confirms the platform's strength. Additionally, NVL72 computing units are already in use on cloud platforms such as CoreWeave, Google Cloud, and Microsoft Azure. Even French AI company Mistral plans to deploy thousands of Vera GPUs in Europe, investing billions in building new computing centers. In short, the industry's leading players are eagerly adopting Vera Rubin, indicating that it has become a standard for the intelligent agent era.
2. Power Efficiency Increased by Tenfold: Solving Power Consumption Challenges
One of the biggest challenges in AI computing is high power consumption. An AI facility may require several megawatts or even gigawatts of electricity, resulting in substantial costs. Vera Rubin significantly improves “power efficiency per watt.” CoreWeave tests show that its performance per watt is ten times that of its predecessor, meaning it can handle more AI tasks with the same amount of power and reduces the cost per generated text or audio output. This is a major breakthrough for AI facilities with limited power resources.
3. Intelligent Hardware Design: Faster Assembly, Water Savings, and Environmental Friendliness
The NVL72 computing units are designed to be highly efficient:
- Faster assembly: There are no cables, fans, or hoses in the chassis, reducing assembly time from several hours to just one minute, which saves time and money for large-scale deployments.
- Water savings: The use of a 45°C liquid cooling system eliminates the need for chillers, saving hundreds of gallons of water per watt annually (equivalent to the annual water consumption of a household). With increasing environmental regulations, such designs make it easier for AI facilities to meet compliance requirements and reduce operating costs.
4. Enhanced Networking Speeds: smoother Operation for AI Clusters
Large AI models require rapid data transfer between chips, and slow networks can impede performance. Vera Rubin's networking improvements are impressive:
- 6th-generation NVLink: Data throughput has doubled, and latency has been reduced by three times, ensuring faster and more reliable communication between chips.
- Spectrum-X Ethernet: The latest switches and network cards provide data transfer bandwidth that is 1.6 times higher than standard Ethernet solutions. Companies like SpaceX and Tesla are already using this technology, indicating its capability to support larger AI clusters for complex tasks such as autonomous driving and large model training.
5. A New $200 Billion Market: NVIDIA Expands Its Business Opportunities
Jensen Huang described Vera Rubin as a platform built for the “era of intelligent agents.” It is now in full production, and virtually all major model developers and cloud providers are preparing to build their infrastructure on it. More importantly, the Vera CPU alone opens up a new market worth $200 billion for NVIDIA. This means that NVIDIA can generate revenue not only from GPUs but also from CPUs and the entire platform ecosystem, further expanding its growth potential.
In Conclusion
The Vera Rubin platform is more than just a hardware upgrade; it represents a comprehensive computing solution tailored for the intelligent agent era. By addressing key challenges in the AI industry, it has attracted leading players and created significant commercial opportunities for NVIDIA. For consumers, this means that future AI applications (such as smarter chatbots and autonomous vehicles) will become more accessible and affordable.