TechCrunch is reporting that Nvidia chief executive Jensen Huang has laid out his case for why the company expects growth of roughly 70 percent in the coming year, pushing back along the way against suggestions that the company's expanding web of commercial relationships has a self-reinforcing, circular quality to it.
To understand why that defense matters as much as the growth forecast itself, it helps to appreciate the position Nvidia now occupies in the technology industry. The company began as a graphics chip maker serving video game enthusiasts and, for a long time, that remained its defining identity. What changed everything was the realization, gradual at first and then explosive, that the same parallel-processing architecture that rendered game environments could train artificial intelligence models with remarkable efficiency. When the deep learning wave broke across the industry in the early part of this decade, Nvidia's GPU hardware was already there to meet it, and competitors scrambled to catch up from a standing start.
The result is that Nvidia does not merely sell components. It sells what amounts to the foundational infrastructure layer on which modern AI is being built. Its chips power the data centers of cloud providers, its CUDA software platform has become the default programming environment for AI researchers, and its networking hardware — acquired through the purchase of Mellanox — knits those chips together inside massive computing clusters. When a technology company wants to build a large language model or a generative AI product, it almost invariably passes through Nvidia's ecosystem at multiple points. That breadth is what Huang means when he or his interviewers speak of having a finger in every pie.
The circularity question is subtler and worth taking seriously. Critics and analysts have noted that some of Nvidia's largest customers are also, in various ways, partners or investors in ventures that then return to buy more Nvidia hardware. Large cloud providers buy chips to rent out AI computing capacity; AI startups raise money from those same cloud providers; the startups use that money partly to rent compute; the compute demand justifies more chip purchases. Huang's insistence that these deals are not circular is, on one level, a reputational defense — circular arrangements can attract regulatory attention and raise questions about whether reported demand is genuine or artificially sustained. His argument, the likely reading of it, is that each transaction reflects real underlying demand from real end users, and that the chain of relationships reflects normal commercial interdependency rather than something more contrived.
Whether that argument fully satisfies skeptics is another matter. The history of technology booms includes periods when capital recycled through ecosystems in ways that inflated apparent demand before a correction arrived. That does not mean the same dynamic is at work here, but it explains why the question keeps surfacing and why Huang felt it necessary to address it publicly.
The 70 percent growth figure, if realized, would be extraordinary for a company of Nvidia's current scale. Growth at that rate is easier to achieve from a small base; sustaining it when revenues are already measured in tens of billions of dollars requires genuine and expanding demand rather than the same customers simply spending more. The breadth of Nvidia's current positioning — across cloud infrastructure, sovereign AI projects in various countries, automotive applications, and the physical AI and robotics space Huang has spoken about frequently — is the structural argument for why such growth could be achievable rather than aspirational.
The consequences of this trajectory fall on several different groups. For competitors, including AMD, Intel, and the custom chip efforts underway at Google, Amazon, and Microsoft, a Nvidia that continues compounding at this rate makes the competitive hill steeper with each passing quarter. Switching costs in AI infrastructure are high, and a developer base trained on CUDA is not easily migrated. For the broader technology sector, Nvidia's fortunes function almost as a proxy for AI investment sentiment; when its numbers disappoint, the ripples spread quickly across the valuations of companies throughout the supply chain and beyond. For regulators, particularly in the United States and Europe, the concentration of AI computing capacity in a single company's hardware is increasingly a matter of policy interest, and growth of this magnitude will only intensify that attention.
What to watch for next is whether the demand signals Huang is pointing to translate into the actual order volumes and revenue that would validate his forecast when Nvidia next reports earnings. Equally important will be any movement from major cloud providers or governments on diversifying their chip suppliers, which would test the stickiness of Nvidia's ecosystem advantage. And the circularity debate is unlikely to go away quietly — if regulators on either side of the Atlantic decide to examine the structure of AI infrastructure investment more closely, Nvidia's commercial relationships will almost certainly be part of the conversation.
