What happened
NVL72 rack production is ramping up with key cloud partners, supported by a global supply chain.
Why it matters
The large-scale adoption of the new architecture by leading cloud providers signals a critical phase in the renewal of global computing capacity, which could determine the pace of artificial intelligence development over the coming year.
NVIDIA announced the start of mass production for its new Vera Rubin architecture, characterized by a transition to gigascale operations. Production of NVL72 format racks is actively scaling, and the first working units are already functioning within partner infrastructure, including CoreWeave, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure.
To support this expansion, an unprecedented logistics network has been deployed, covering more than 350 factory sites in 30 countries. According to the manufacturer, this is the largest and most mature rack-scale supply chain ever assembled to meet growing demand for compute power.
The launch aims to achieve maximum performance per watt of energy and lower token costs for partners worldwide. The scale of deployment indicates infrastructure readiness for the immediate processing of significant volumes of computational tasks.
Facts
- NVIDIA has begun gigascale production of the Vera Rubin architecture.
- Production of Vera Rubin NVL72 racks is ramping up.
- Racks are already operating with partners: CoreWeave, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure.
- The supply chain includes more than 350 factory sites in 30 countries.
- NVIDIA calls this supply chain the largest and most mature of its kind.
Context
Information is based exclusively on the meta-description of the official NVIDIA blog from July 2026. The report contains no independent confirmation from the mentioned partners or third-party analysts.
What remains unknown
- What are the actual energy efficiency and token cost figures compared to previous generations?
- What are the real production volumes and timelines for full line utilization?
- Have partners confirmed their data centers' readiness for mass integration of the new racks?
AI analysis
The emphasis on the number of factories and countries indicates that NVIDIA's primary challenge is not chip design, but the physical scaling of production and logistics. Mentioning specific names of major cloud providers serves as a signal to the market that demand is already secured by contracts, reducing the risks of idle new capacity.
Strategic AI conclusion
A likely consequence will be intensified competition for access to next-generation computing resources among players not included in the initial partner list. The next observable signal will be cloud providers' financial reports on capital expenditures. A key uncertainty remains the ability of the global power grid to support the sharp rise in energy consumption by new racks.