Nvidia's Vera Rubin Enters Production Amid Spiking Demand and Looming Price Hikes

Nvidia has confirmed that its Vera Rubin AI computing platform is in full production and scheduled for partner availability in the second half of 2026. [2] The architecture's flagship NVL72 system packs 72 Rubin GPUs and 36 Vera CPUs into a single rack to deliver 3.6 exaflops of NVFP4 inference compute and 2.5 exaflops of training compute. [2] A fully scaled Vera Rubin POD can stretch to 40 racks, totaling 1,152 Rubin GPUs and roughly 60 exaflops of NVFP4 compute. [2] Nvidia claims the Rubin architecture delivers five times the inference performance of current Blackwell systems at the rack level. [2]

Dell Technologies is already utilizing the Vera Rubin NVL72 to power its AI Factory, which is delivering ten times lower cost per token for AI inference. [1] Currently, 5,000 enterprises including Lilly, Samsung, and Honeywell are leveraging Dell AI Factories powered by Nvidia to move beyond pilot programs and into full-scale deployments. [1] Dell is also introducing the PowerEdge XE9880L, XE9885L, and XE9882L servers, which are built on the Nvidia HGX Rubin NVL8 and capable of supporting up to 144 GPUs per rack with fully liquid cooling. [1]

The financial impact of this hardware cycle is visible in Nvidia's latest earnings, where the company reported data center revenue hitting $75 billion, representing a 92 percent increase. [6] Nvidia's networking revenue alone nearly tripled year over year. [6] To complement its GPU business, Nvidia is now forecasting $20 billion in CPU sales this year. [4] The company also authorized an $80 billion share buyback and increased its dividend to 25 cents per share. [6]

Nvidia CEO Jensen Huang characterized the surge in demand for artificial intelligence processing power as going parabolic. [1] Huang told investors the company anticipates $1 trillion in combined Blackwell and Rubin revenue from 2025 through calendar 2027. [6] Token consumption is projected to grow by 3400 percent alongside a broader $3-4 trillion AI infrastructure investment anticipated in the coming years. [1]

Alongside this disclosure, CNBC also noted that various Nvidia customers are reportedly being warned about upcoming AI-related price hikes. [8] Some of Nvidia's top customers have reportedly been informed that prices for systems including Vera Rubin and Grace Blackwell will increase by 15 percent or more starting in early 2027. [7]

While Nvidia continues to dominate the AI infrastructure market with record data center revenues and a sprawling Blackwell and Rubin pipeline, competitors are securing significant deployments that could shift the market balance. AMD recently posted $11.54 billion in record revenue, and leading foundation model builder Anthropic committed to utilizing up to two gigawatts of AMD's MI450 series GPUs starting in 2027. Nvidia's planned 15 percent system price increases in early 2027 could test customer loyalty precisely as AMD ramps up its alternative hardware ecosystem and validates its second-source strategy. Furthermore, Nvidia's aggressive push into standalone CPU sales represents a significant expansion of its total addressable market beyond core acceleration. What the sources don't address: whether Nvidia's upcoming hardware price hikes will prompt major cloud providers to more aggressively shift their inference workloads to custom in-house silicon rather than paying a premium for Vera Rubin.