Based on the latest market reports and the GTC keynote, the stock movement of NVIDIA (NASDAQ: NVDA) is being driven by a strategic focus towards AI Networking and Inference. During the NVIDIA GTC 2026 keynote on March 16, 2026, CEO Jensen Huang projected a massive surge in cumulative orders for Blackwell and Vera Rubin architectures through 2027 to reach $1 trillion. By establishing the networking hardware for AI factories as the primary revenue engine for the company, NVIDIA is becoming the central architecture of AI infrastructure and is currently the world’s largest networking company by revenue.
While the reveal of the ‘Vera Rubin’ GPU was a major step into this realm, the silent networking breakout of NVIDIA is becoming the foundational part of the company’s explosive growth. As per the latest 2026 Q4 earnings release of NVIDIA, the company reported a record revenue of $68.1 billion, with 20% surge from Q3 and a 73% surge from the previous year. The quarterly networking revenue has reached $10.98 billion, making a 263% year-over-year gain, and the total company revenue has grown by 65% to $215.9 billion.
“Computing demand is growing exponentially – the agentic AI inflection point has arrived. Grace Blackwell with NVLink is the king of inference today – delivering an order-of-magnitude lower cost per token – and Vera Rubin will extend that leadership even further,” said Jensen Huang, the founder and CEO of NVIDIA. He has officially declared the transition of the company from a ‘chip company’ to the world’s largest networking provider.
The AI Factory Blueprint and the Vera Rubin Pivot
NVIDIA is moving beyond the standalone GPUs to the Vera Rubin NVL72, a rack-scale system integrating Rubin GPUs, Vera CPUs, NVLink 6, and Spectrum-6 networking, which are designed to address the ‘Inference Inflection’ by enabling agentic AI at scale. They can eliminate the bottlenecks from computer networking by enabling massive deployment of trillion-parameter models.
NVIDIA controls the entire stack, which includes BlueField-4 DPU and proprietary cabling, to execute an ‘Extreme Co-Design’ infrastructure that can optimize the inference. This next-gen infrastructure can facilitate 10x higher inference throughput per watt, while the Vera CPUs handle Agentic AI, and BlueField-4 controls the infrastructure overhead. By tailoring an entire stack to navigate Agentic AI, this strategic move is creating a pivotal competitive advantage for the company over the traditional networking chip providers.
Market Outlook: The $1 trillion Pipeline
Based on the updated guidance by the CEO at the global AI conference, the previously provided $500 billion demand estimate for the period through 2026 has doubled. Huang claims the fundamental shift towards Agentic AI and ‘inference supercycle’, which brings explosive demand for global AI infrastructure, as the primary drivers for this growth.
This projection, which is considered the largest order book in semiconductor history, mirrors the money flow under the impact of the booming AI business across the globe. While the figure specifically represents the Blackwell and Vera Rubin platforms, including related networking, it excludes newer products like Groq LPUs and standalone CPUs. While the Vera Rubin Architecture will be entering production in 2026 with a 3.5x faster training capacity and 5x faster inference capacity than Blackwell, its higher-performance variant, the Vera Rubin Ultra, is targeted for release in 2027.
That being said, NVIDIA is no longer a cyclical semiconductor stock, but a foundational utility for the ‘AI Token Economy.’ Based on the revenue projected for Q1 2027, the financial outlook of the company gives no indication of a cooling off. Around 60% of NVIDIA’s business remains backed by the top five hyperscalers, such as Microsoft, AWS, and Google, although a demand spike can be observed across startups and enterprises. For investors, the ‘Networking’ line in the balance sheet is becoming as important as the GPU itself. Which means as long as AI agents require bandwidth to operate, NVIDIA’s infrastructure will remain at the top in the industry.
