To The Moon

NVIDIA made a significant impact at today's GTC conference by announcing three key updates, effectively dispelling market concerns over its product roadmap and demonstrating an expansion of its business from GPUs into CPUs and the consumer AI PC market.

CEO Jensen Huang announced that the flagship AI platform, Vera Rubin, has officially entered mass production, alleviating investor worries about potential delays. Concurrently, NVIDIA introduced the Vera CPU, designed specifically for agentic AI workloads, which reportedly delivers significantly superior performance compared to x86 competitors. The company also collaborated with Microsoft to launch the new ARM-based Windows PC, RTX Spark.

According to market analysis, Citi Research analyst Atif Malik characterized these developments as "positive" signals. A particularly noteworthy point is that the launch of the Vera CPU suggests the historical sales content value ratio of CPU to GPU is evolving from 1:2 towards 1:1, opening new revenue growth avenues for NVIDIA.

Citi maintains a Buy rating on NVIDIA with a price target of $300, implying an approximate 42% upside from the current price of $211.14.

**NVIDIA (NVDA)**

The confirmation of Vera Rubin's mass production was a highlight of the announcements. Citi Research noted that the market had harbored concerns about the platform's timely delivery, which were effectively addressed by Huang's public confirmation.

The Vera Rubin platform, comprising five dedicated racks that function as a unified AI supercomputer, integrates the Vera Rubin NVL72 GPU system, Vera CPU, BlueField-4 STX storage, and Spectrum-6 networking. Compared to the previous Blackwell generation, the platform offers up to 10x agent throughput at scale. Major server manufacturers and supply chain partners have begun batch manufacturing Vera Rubin-based systems for AI labs, cloud service providers, and hyperscale data center operators.

Also entering mass production is the NVIDIA Spectrum-X Ethernet photonic interconnect technology. By combining co-packaged optics with Spectrum-X switches, it enables the networking of million-GPU-scale AI factories, offering 5x greater energy efficiency, 5x longer AI runtimes, and 1.3x faster deployment speeds compared to networks using traditional transceivers. CoreWeave, Microsoft, and OCI are among the first ecosystem partners.

**NVIDIA (NVDA)**

The strategic elevation of NVIDIA's CPU business came as a surprise to many analysts.

Huang positioned the Vera CPU as the core bottleneck for the agentic AI era, stating, "In the age of AI agents, the CPU has become the bottleneck for GPU utilization." The Vera CPU is built for agentic AI workloads, delivering 3x the performance of x86 competitors in SQL operations, 6x in data processing, and approximately 1.8x overall performance in common agent tools like Python, code analysis, and compilation.

More significant is the changing CPU-to-GPU sales ratio. Citi Research points out that historically, NVIDIA sold one CPU for every two GPUs. With the rise of agentic AI demand, this content ratio is approaching 1:1. NVIDIA has already launched a standalone CPU rack product comprising 256 CPUs, which also serves a memory expansion function, further enhancing its commercial value. Vera CPU has secured Anthropic and OpenAI as early-adopter customers.

Citi Research believes this development is substantively meaningful for increasing NVIDIA's overall system value, with CPU business revenue potentially exceeding prior market expectations.

**NVIDIA (NVDA)**

In the personal computing arena, NVIDIA, in collaboration with Microsoft, launched the RTX Spark superchip, extending agentic AI capabilities to consumer Windows devices.

RTX Spark utilizes an ARM architecture, connecting a Blackwell RTX GPU to a 20-core Grace CPU via NVIDIA NVLink C2C. It delivers 1 petaflop of AI performance with up to 128GB of unified memory, targeting thin-and-light laptops with all-day battery life and highly efficient small-form-factor desktops.

In terms of capability, RTX Spark can render 3D scenes over 90GB, edit 12K 4:2:2 video, run large language models with 120 billion parameters and million-token context windows locally, and run AAA games at 1440p resolution with over 100 frames per second. NVIDIA stated the product incorporates three decades of core technology, including CUDA, RTX, DLSS, FP4, TensorRT, and OptiX.

Citi Research characterized this launch as the dawn of a "new PC era," believing RTX Spark integrates AI creation, gaming, and personal agent capabilities into a single consumer device.