Nvidia Opens Vera Rubin Rack Design, Setting Stage for AI Data Center Standards Battle

Nvidia Opens Vera Rubin Rack Design, Setting Stage for AI Data Center Standards Battle

TREE NEWS reports: At the Open Compute Project (OCP) Global Summit in San Jose, California, Nvidia Vice President Ian Buck announced that the company will open the server rack and compute tray design blueprints for its next-generation Vera Rubin AI platform to external partners. The move, which includes common specifications for power delivery, liquid cooling, and network connectivity, aims to lower development barriers for server manufacturers and shorten data center construction cycles and costs. Nvidia also unveiled a dedicated CPU rack system based on its in-house Vera CPU, targeting AI agent workloads that require concurrent management of multiple programs.

Strategic Intent: Extending Influence Beyond GPUs

Industry analysts see a clear strategic motive behind the open design initiative. As the AI agent era arrives, infrastructure demand is expanding from GPUs to CPUs and networking equipment. By opening its design specifications, Nvidia is racing to build an AI infrastructure ecosystem centered on itself, locking in influence before industry standards solidify. The company is seeking to extend its dominance from GPU market share to broader AI data center architecture.

800V DC Power Architecture Addresses Rack Power Bottlenecks

In the power architecture, Nvidia proposes raising rack DC voltage to 800 volts. Under the blueprint, conversion units inside data centers will convert external AC power to 800V DC before delivering it to server rack entrances. Previously, data centers typically used 48 to 54 volts DC internally. However, as single-rack power consumption climbs to approximately 1 megawatt, the current intensity at low voltage becomes too high, creating a physical bottleneck for scaling. The high-voltage 800V DC solution is seen as a key technological path for AI data centers to cope with rising energy density.

Dedicated CPU Rack Targets AI Agent Workloads

Nvidia simultaneously released a dedicated rack design specification based on its self-developed Vera CPU. The existing Vera Rubin NVL72 system combines 72 GPUs with 36 CPUs, primarily for large-scale computing tasks. The new Vera CPU rack can carry up to 256 Vera CPUs, positioned for scenarios involving concurrent operation and management of multiple programs, focusing on pure CPU processing power. Joe DeLaere, who leads AI infrastructure solutions at Nvidia, stated that the real bottleneck in the industry is whether AI factories can be manufactured and deployed quickly enough. Opening design specifications is interpreted as Nvidia’s proactive move to accelerate AI data center construction and a strategic step to secure话语权 during the standard formation period.

Market Implications: A Trillion-Dollar Equipment Market Accelerates

The scale of the data center equipment market provides ample competitive motivation for this standards battle. Research firm IDC estimates that spending on equipment compliant with OCP technical standards will reach approximately $800 billion by 2029, about 2.7 times its forecast of $295 billion from a year earlier. By 2030, related spending will further rise to $892 billion, approximately 5.8 times the actual 2025 spending of $154 billion. In this rapidly expanding market, whoever’s design specifications become the industry benchmark will gain first-mover advantages in supply chains and ecosystems. Nvidia’s action is a direct response to this logic.

Key Takeaways for Investors

• Nvidia’s ecosystem lock-in intensifies: By opening rack designs, Nvidia is not ceding control but rather embedding its specifications deeper into the AI data center stack. This could strengthen its long-term pricing power and competitive moat beyond GPUs, benefiting its stock and partners aligned with its standards.
• Power and cooling infrastructure winners: The shift to 800V DC and liquid cooling creates opportunities for companies specializing in high-voltage power conversion, liquid cooling systems, and advanced thermal management. Investors should watch suppliers of these components.
• Server manufacturers face both opportunity and risk: Lower R&D barriers could accelerate time-to-market for server makers, but those that fail to adopt Nvidia’s standards may find themselves marginalized. Dell, Supermicro, and others in the OCP ecosystem may benefit.
• CPU competition heats up: Nvidia’s Vera CPU rack directly challenges Intel and AMD in data center CPUs, potentially reshaping market share in the server processor space.
• Broader AI infrastructure spending boom: IDC’s upward revision of OCP-compliant equipment spending underscores the massive capital inflows into AI infrastructure. This supports demand for semiconductors, networking gear, and data center REITs.