At the Goldman Sachs Communacopia + Technology Conference on September 10, 2026, Nvidia CEO Jensen Huang addressed skepticism regarding the sustainability of the AI boom by reiterating a 70% year-over-year revenue growth forecast for the company’s 2028 fiscal year. The projection highlights a shift in Nvidia’s business from selling individual components to architecting massive, integrated “AI factories.”
Huang framed this growth as the result of a fundamental change in the scale of computing. While Nvidia was once known for consumer-grade graphics cards, it now produces integrated GPU systems that cost approximately $8.5 million per unit and weigh two tons. These systems are so large they require dedicated cargo planes for transport.
Defending the Ecosystem Against ‘Circularity’ Critics
The aggressive growth forecast comes amid scrutiny regarding Nvidia’s investment strategy. Critics have raised concerns over “circular financing,” suggesting that Nvidia may be artificially inflating demand by investing in startups that then use those funds to purchase Nvidia hardware. Huang dismissed these allegations, characterized the investments as a method for building technology distribution channels rather than generating revenue. He argued that Nvidia’s venture activities—including its significant investment in Hugging Face—are immaterial to the company’s total revenue and are intended to expand the broader AI ecosystem.
The demand side of the ledger remains robust, headlined by a massive commitment from Amazon. The cloud giant has reportedly signed an agreement to purchase 2 million Nvidia GPUs for delivery across 2027 and 2028, providing a multi-year floor for Nvidia’s data center business.
Cybersecurity: The Next Billion-Dollar Vertical
While Large Language Models (LLMs) like GPT-4 drove the initial wave of Nvidia’s valuation, Huang identified cybersecurity as the “next major” market for AI growth. Nvidia is currently partnering with firms such as Cisco and CrowdStrike to integrate AI directly into defense workflows. The goal is to move beyond software-based security toward hardware-accelerated threat detection that can process data at the speed of modern networks.
This shift coincides with a significant increase in the “average selling price” of Nvidia’s flagship hardware. The upcoming Vera Rubin platform is expected to be priced at approximately $40,000 per GPU. This represents a steep climb from the $25,000 price tag associated with the Blackwell architecture, reflecting the increased complexity and performance requirements of the next generation of AI training.
The Architecture Shift to Vera Rubin
The transition to the Vera Rubin architecture is central to Nvidia’s 70% growth target. By moving from a semiconductor supplier to a provider of full-scale industrial systems, Nvidia is positioning itself to capture a larger share of the total capital expenditure being spent on global data center upgrades.
Huang noted that the “AI factory” model requires a different logistical approach than traditional tech hardware, as the systems are so massive and power-hungry. This shift is intended to make Nvidia’s hardware an indispensable part of the world’s industrial infrastructure.
