Nvidia Reassures on Product Roadmap and Broadening AI Demand, Citi Reports

Nvidia Corp. has reaffirmed that its AI product roadmap is on schedule and that demand for its chips is expanding beyond its core hyperscaler customers, addressing key questions from the investment community. The update was detailed in a research note from Citi following discussions with the chipmaker's investor relations team.

Demand Broadens Beyond Cloud Giants

According to the note, Nvidia is seeing demand momentum build beyond the initial wave of AI infrastructure deployments by large cloud providers. The company highlighted that AI laboratories, sovereign customers, and on-premise enterprise deployments are expected to become a larger share of the market as AI adoption accelerates.

Nvidia also clarified its position in the AI model ecosystem, stating that both open-source and proprietary models are important. The company's in-house models, such as Nemotron and Cosmos, are designed to support enterprise and government adoption rather than to compete directly with leading frontier-model developers, Citi reported.

Roadmap and Technology on Track

Investors have been closely monitoring for any disruptions to Nvidia's product cadence, but the company stated its Rubin Ultra roadmap remains "fully intact." The NVLink domain architecture, a key interconnect technology, has not changed since it was unveiled at Computex.

Ad

Nvidia also provided an update on its networking technology, noting that co-packaged optics (CPO) are already in production with its Spectrum-X platform and seeing high customer adoption. The company added that beginning with its Feynman platform in calendar 2028, customers will have the choice between NVLink implementations using either CPO or copper interconnects.

Financial Outlook and Efficiency Gains

The chipmaker reiterated several key financial targets, providing clarity on its capital and operational plans:

• It maintained its gross margin guidance in the mid-70% range.
• The company is targeting a return of 50% of cash flow to shareholders this year, with buybacks potentially increasing over time.
• A recent $25 billion debt offering was intended to provide greater financial flexibility.

Management also clarified recent comments from CEO Jensen Huang regarding energy efficiency. The projection that $100 billion of compute could eventually require just one gigawatt of power reflects expected efficiency improvements in future products, not current infrastructure costs. For context, Nvidia estimates its current systems generate roughly $30 billion to $40 billion of compute revenue per gigawatt.