Citigroup: Nvidia's revenue per GW may rise from $18 billion to $40 billion, with the value of AI infrastructure continuing to increase.

Nvidia is evolving from a GPU chip company into a full-stack platform company covering computing power, networking, security, and AI infrastructure, as the AI wave expands both the scale of demand and the value of individual projects.

Citigroup recently released a report maintaining its "buy" rating on Nvidia with a target price of $315. The report points out that Nvidia's additional $150 billion in stock repurchase authorization brings its total repurchase capacity to $235 billion as of FY28, reflecting the company's confidence in the continued growth of its AI-driven free cash flow.

At the same time, NVIDIA is extending into AI security, compliance, governance, and complete AI factory infrastructure, and the rapid adoption of open weight models is also continuously driving up the demand for token consumption and underlying computing power.

More importantly, Nvidia is increasing the value of each AI infrastructure deployment. The company expects the revenue opportunity per GW to increase from $18 billion in the Hopper era to $25 billion in the Blackwell era, and further to $40 billion in the Rubin era.

$150 billion buyback: Cash flow confidence behind the massive authorization

Following the addition of $150 billion to its stock buyback authorization, Nvidia's total buyback capacity reached $235 billion as of FY28. With the company's market capitalization at $5.52 trillion, this scale of buybacks has become an important component of its capital allocation strategy.

Citigroup believes that the further expansion of the share buyback program reflects Nvidia's confidence in future free cash flow growth. With the continued expansion of demand for AI infrastructure, the company expects its profitability and cash generation capabilities to remain strong, thus enabling it to maintain large-scale shareholder returns while continuing to increase investment in AI.

In other words, Nvidia did not make a trade-off between growth investment and capital returns. On the one hand, the company continues to invest in AI infrastructure, software, and emerging applications; on the other hand, it also retains ample room for share buybacks.

Extending from GPUs to AI Security: Business Boundaries Continue to Expand

NVIDIA has launched the Open Agent Safety Platform, marking its entry into the security, compliance, and governance aspects of AI agents. The platform includes the OpenShell open-source runtime and the Sentry hardware monitoring system running on the BlueField-4 DPU, enabling policy control over AI agents and the detection and blocking of unauthorized behavior.

As AI agents are integrated into enterprise-critical workflows, the demand for security and control capabilities is increasing. NVIDIA is thus expanding its service offerings by extending beyond underlying computing to the security layer required for AI applications to run.

At the same time, the open model ecosystem is expanding rapidly. Currently, open models account for approximately 75% of token generation, compared to about 40% a year ago; overall token consumption has increased 25 times year-on-year. Nvidia believes that the increased openness of models is driving the popularization of AI applications, while the growth in token demand is further boosting the demand for underlying computing power.

The company continues to invest in areas such as language AI, physical AI, robotics, and autonomous driving, and plans to acquire Hugging Face to strengthen its connection with the developer ecosystem.

Revenue potential per GW increased from $18 billion to $40 billion.

Nvidia predicts that the revenue opportunity per GW has increased from $18 billion in the Hopper era to $25 billion in the Blackwell era, and will reach $40 billion in the Rubin era.

This growth stems not only from improved GPU performance but also from the integration of CPUs, LPUs, network devices, and AI factory infrastructure into overall solutions. This allows NVIDIA to cover more aspects of AI infrastructure deployment, increasing revenue per project.

Order data also shows that demand for AI infrastructure continues to expand. Nvidia disclosed that a cutting-edge AI lab has already signed a direct contract for 2.6GW of Nvidia AI infrastructure, with the equipment to be delivered before 2028; in addition, the combined value of indirect contracts with multiple cloud service providers (CSPs) and emerging cloud vendors has exceeded $180 billion.

This means that Nvidia's growth comes not only from the expansion of its AI infrastructure deployments, but also from the company covering more products and processes in individual projects. As Hopper iterates towards Blackwell and Rubin, the revenue opportunities per GW are also continuously increasing.

Risk warning and disclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.