Nvidia CEO: Quarterly Revenue Nears $100 Billion as Growth Accelerates; Denies Delay of Rubin Ultra

Recently, Nvidia CEO Jensen Huang stated at a non-deal roadshow that despite the company's quarterly revenue approaching the $100 billion mark, its growth rate has not peaked and is actually continuing to accelerate.

Furthermore, addressing recent market concerns, Nvidia's management directly responded to three major doubts: intensifying competition from ASICs, potential delays in Rubin products, and whether AI computing demand has peaked. They painted a picture of AI computing demand continuously expanding from traditional cloud giants to new sectors such as Sovereign AI and Industrial AI.

Debunking Rubin Ultra Delay Rumors: Merely Architectural Optimization, Shipping Schedule Unchanged

Huang explicitly denied recent market rumors suggesting that the next-generation flagship architecture, Rubin Ultra, might be delayed until 2028, confirming that the product will still officially ship next year as planned.

Nvidia further explained that while the rack design for some Rubin systems is indeed being adjusted—with the original Kyber rack solution being replaced by a new design to support larger-scale AI computing clusters—this is strictly an optimization at the system architecture level. Key technical modules, including the 800V high-voltage power supply system and inter-rack optical interconnects, are progressing as scheduled. The R&D and delivery pace of the next-generation products has not been substantially affected.

Three Growth Engines Firing: Continued Diversification of Customer Base

Addressing market concerns over Nvidia's revenue being overly reliant on a few top-tier clients, management outlined three core growth directions for the future during the meeting:

First, AI Labs. This segment currently accounts for approximately 20% of Nvidia's overall demand. Beyond the continued deep integration with top-tier clients, frontier model companies like Anthropic, which previously favored ASIC chips, are now rapidly increasing their GPU deployments.

Second, Traditional Hyperscale Cloud Service Providers. Microsoft, Meta, Amazon, and Google collectively contribute about half of Nvidia's revenue. Although the expansion of these clients is constrained by power supply, land approvals, and data center construction cycles, Nvidia's product offerings to them are expanding from standalone GPUs to include CPUs and networking equipment, further broadening its revenue boundaries.

Third, Emerging AI Cloud Platforms, Sovereign AI, and Enterprise Customers. Driven by geopolitical conflicts and data security needs, governments worldwide are actively building domestic AI computing infrastructure. These projects tend to favor highly integrated, full-stack solutions and face less competitive pressure from custom ASICs, making them the fastest-growing market segment.

CPU Business Targets $20 Billion as Vera Enters the General-Purpose Server Market

Beyond its absolute dominance in GPUs, Nvidia once again emphasized the strategic importance of its CPU business. The company expects its CPU revenue to reach approximately $20 billion this fiscal year.

Notably, nearly half of this revenue may come from standalone CPU racks. This signifies that the next-generation Vera CPU is no longer confined to serving merely as a management node within GPU servers. Leveraging an architecture optimized for single-threaded workloads, it is officially entering the massive general-purpose server market. As Nvidia transitions from a standalone GPU supplier to a full-stack AI infrastructure platform encompassing GPUs, CPUs, and network interconnects, its total addressable market continues to expand.