Table of Contents

• You might want to know


• Main Topic


• Key Insights Table


• Afterwards...

You might want to know

1. How does Nvidia’s hardware and ecosystem strategy create what Jensen Huang calls visibility into AI demand?

2. Could competitors, in-house silicon efforts by cloud providers, or changing infrastructure efficiency undercut Nvidia’s forecast?

Main Topic

At the Goldman Sachs Communicopia + Technology conference, Nvidia’s founder and CEO, Jensen Huang, laid out a forceful argument for why the company expects sustained, rapid growth in the AI era — including a forecast of roughly 70% year-over-year revenue growth next year. His case rests on multiple interlocking elements: the scale and specialization of contemporary AI hardware deployments, Nvidia’s wide-reaching role across the AI stack, and the company’s intimate visibility into the global build-out of AI infrastructure.

Huang began by reframing how people think about Nvidia’s products. The company no longer sells modest discrete chips to upgraders in the PC gaming market; instead, it delivers complex, massively scaled systems. To illustrate, he contrasted a historical perception of a $399 GPU with the modern, hyperscale systems Nvidia supplies: configurations that can cost millions per unit and include thousands of interlinked components, specialized interconnects, and very large power requirements. In Huang’s words, a single modern GPU deployment can represent millions of dollars in hardware and infrastructure.

This shift in unit economics matters because it changes how demand translates into revenue. Large, integrated systems are not one-off retail purchases; they are multi-million-dollar orders that scale quickly when AI projects expand. Huang cited a specific product mix example: a computer system combining dozens of CPUs and GPUs that is experiencing strong month-over-month growth. He pointed to a device configuration containing 36 Grace CPUs paired with 72 Blackwell GPUs and noted that sales for that class of system were growing at roughly 27% month over month. Such growth rates on large-ticket systems can compound quickly into very large revenue increases.

Beyond unit economics, Huang emphasized Nvidia’s embedded position within the AI ecosystem. He described Nvidia as a foundational platform used across an array of AI initiatives — from hyperscalers and cloud providers to AI-native startups and research labs. That breadth of engagement means Nvidia supplies chips and systems to many of the same organizations that are building and training the most demanding AI models, including those from large AI labs and public cloud operators. By supplying the compute these actors need, Nvidia acquires a panoramic view of where, when, and how AI capacity is being deployed. Huang framed this visibility as a strategic advantage: "We’re tracking every single gigawatt of land, power, shell around the world," he said, referring to data-center shells and the power allocations flowing into AI projects. This level of information helps Nvidia forecast demand with a degree of granularity few suppliers can match.

Another pillar of Huang’s rationale was Nvidia’s commercial approach and capital partnerships. The company has entered into financing and investment arrangements with customers and partners, which has drawn scrutiny due to historical precedents where circular capital flows masked underlying weaknesses in demand. Huang addressed this critique directly and with a mixture of humor and business clarity: while Nvidia may place equity or capital into partner companies, the investments typically follow demonstrated commercial contracts and tangible purchasing commitments. He asserted that the company sees tens of billions of dollars in real, revenue-generating contracts that underlie its exposure to the market — framing the investments as catalytic rather than speculative.

Huang’s bullishness also rests on the diversity of workloads and models that rely on Nvidia hardware. He asserted that "every model" and "every single lab" can use Nvidia’s platform, highlighting the company’s compatibility with open-weight models as well as proprietary systems from major AI labs. This cross-cutting compatibility strengthens the company’s position as a default choice for many organizations building AI capabilities, reinforcing recurring demand across multiple sectors.

However, Huang and his company do not ignore the competitive landscape. He acknowledged that hyperscalers — including Amazon, Microsoft, and Google — are developing their own silicon, and specialized startups and public competitors are also targeting AI hardware. The fundamental question is whether those efforts will scale quickly enough to meaningfully reduce Nvidia’s market share. Huang argued that building competitive parity requires not only designing chips but also creating the complete packaging, interconnects, cooling, software, and supply-chain orchestration that large-scale AI deployments demand. That complexity acts as a barrier to rapid displacement and helps explain why Nvidia often remains central even as others develop in-house or alternative solutions.

There are countervailing forces that could temper Nvidia’s growth over time. As AI workloads and infrastructure mature, organizations may optimize for cost efficiency and token usage, reducing the marginal compute demand per model. Cloud providers may internalize more of their compute stack, and new architectures could shift performance-per-watt dynamics. Still, based on the current trajectory of large-scale model training, the multiplicative effect of big-ticket system sales, and Nvidia’s insight into global data center expansion plans, Huang projects a continuation of exceptionally strong growth into the next fiscal year.

In short, Nvidia’s argument for a potential 70% revenue increase rests on four linked claims: the dramatic change in unit economics for modern AI systems, Nvidia’s pervasive role across AI model development, the company’s granular visibility into data-center and power deployments, and commercial arrangements that convert demand signals into committed purchases. Taken together, these elements form the backbone of Huang’s confident forecast, while acknowledging that longer-term dynamics in efficiency, competition, and architectural innovation will shape how that story evolves.

Key Insights Table

Afterwards...

Looking forward, Nvidia’s near-term prospects appear robust: the company is deeply entwined with the organizations driving AI development and benefits from outsized revenue per deployment. Yet the technology sector’s history warns that lead positions invite competition and innovation. Over the medium term, watch for shifts in infrastructure efficiency, cloud providers’ strategic silicon decisions, and new hardware architectures. If Huawei, hyperscalers, startups, or open-source approaches materially improve cost or performance, Nvidia’s dominance may face pressure — but its current scale, comprehensive ecosystem, and visibility into global AI investments give it a strong runway for continued growth in the immediate future.

Ultimately, Huang’s forecast is a testable thesis: monitor quarterly order trends, reported contract backlogs, and disclosed partnerships to see whether the projected 70% growth manifests or if market and technological changes alter the trajectory. For now, Nvidia’s position in the AI value chain makes its optimistic projection plausible rather than purely rhetorical.