Key Highlights
- Azure claims the distinction of being the first cloud platform to validate Nvidia’s Vera Rubin NVL72 infrastructure
- Satya Nadella, Microsoft’s CEO, shared the news via X (previously known as Twitter) late last week
- A single NVL72 rack provides up to 3.6 exaflops of computational power — a fivefold increase over GB200 configurations
- The system integrates 72 GPUs with 36 CPUs through sixth-generation NVLink technology, achieving 260TB/s connectivity
- Competitors including Amazon, Google, CoreWeave, Nebius, and Oracle plan Rubin deployments throughout 2026
In a significant development for cloud computing infrastructure, Microsoft Azure has secured its position as the first major cloud service provider to validate Nvidia’s cutting-edge Vera Rubin NVL72 system. The announcement came directly from CEO Satya Nadella on Friday via his X social media account, where he characterized it as “another big step in building the next generation of AI infrastructure.”
Nvidia’s Vera Rubin NVL72 represents a rack-scale computing platform that integrates 72 Rubin GPU units alongside 36 Arm-based Vera CPU processors within a unified architecture. These graphics processing units interconnect through sixth-generation NVLink fabric technology, achieving bandwidth speeds of 260 terabytes per second.
The computational leap is substantial. A single NVL72 rack configuration can deliver up to 3.6 exaflops of processing capability — approximately five times the performance of GB200-based predecessor systems currently being phased out.
Rani Borkar, who serves as Microsoft’s President of Azure Hardware Systems, emphasized the extensive preparation behind this achievement. “Microsoft has years of market-proven experience in designing and deploying scalable AI infrastructure that evolves with every major advancement of AI technology,” Borkar explained.
The critical element is collaborative development. Microsoft indicates it has maintained a multi-year partnership with Nvidia spanning interconnect technology, memory architecture, thermal management, packaging solutions, and rack-scale design. This deep collaboration enables Rubin hardware to integrate seamlessly into Azure’s current infrastructure without requiring architectural overhauls.
Strategic Infrastructure Planning Pays Off
Azure’s data center facilities, including major installations in Wisconsin and Atlanta, were purpose-built to accommodate the elevated power requirements and liquid-cooling specifications that NVL72 racks demand. Such forward-looking infrastructure development requires extensive lead time.
Borkar verified that Azure’s “superfactories” were constructed with these systems in mind. “Rubin integrates directly into Azure’s platform without rework,” she noted, highlighting the multi-year preparation underlying what appears as a straightforward first-adopter announcement.
The tech giant undertook comprehensive redesigns of power distribution and liquid-cooling infrastructure across numerous facilities to support the heightened watt density these advanced racks generate. That strategic investment now delivers competitive advantage through validated hardware while competitors remain in queue.
A consortium led by BlackRock, with participation from Microsoft and Nvidia, recently initiated a $40 billion acquisition of Aligned Data Centers, strategically positioning for expanded global infrastructure capacity in anticipation of this next-generation hardware rollout.
Competition Follows Behind in Timeline
While Microsoft claims first-mover status, the competitive landscape will shift soon. Amazon Web Services, Google, CoreWeave, Nebius, and Oracle have all committed to Vera Rubin system deployments — with most targeting the latter half of 2026 for implementation.
Financial analysts at Bernstein have highlighted Microsoft’s “first-to-validate” achievement as evidence of its comprehensive cloud and SaaS operational advantages, quantifying this through what they term a “Rule of 37.3%” performance superiority framework.
MSFT shares declined 1.57% while NVDA dropped 1.58% on the announcement date, tracking broader market weakness rather than indicating negative sentiment toward the validation news.
Rubin Ultra, representing the subsequent evolution of this platform architecture, is scheduled for release in 2027.


