Nvidia's edge is expanding beyond the GPU itself
Nvidia's new Vera Rubin system shows the company's advantage in AI now extends beyond GPUs to the complex infrastructure surrounding them. Investors are reassessing the company's outlook following Wednesday's earnings report.
For years, the story around Nvidia was straightforward: the company was the sole source of cutting-edge GPUs, and it profited enormously as the AI industry scaled up. More recently, hyperscalers like Amazon and Google have begun building their own chips, raising questions about how durable Nvidia's lead really is. After a tenfold increase in market capitalization between early 2023 and mid-2025, Nvidia's share growth slowed over the past year amid concerns about GPU competition.
Following the company's earnings report on Wednesday, a new narrative is emerging: Nvidia's advantage extends well beyond the GPU itself. As AI computing scales toward the gigawatt level, coordinating data flow across massive deployments has become increasingly complex. Nvidia has built much of the specialized hardware needed to manage this complexity, giving it a strong position in the systems surrounding its chips.
Beyond the chip
The company is rolling out its Vera Rubin architecture, which combines the Rubin GPU with additional components — the Vera CPU, the Groq 3 LPX inference accelerator, and dedicated storage and networking racks. According to Jason Hardy, Nvidia's VP of storage technology, the Vera CPU addresses a core bottleneck: memory capacity in any single server is limited, so ensuring data reaches the GPU at the right moment is critical. Hardy said the Vera CPU has enabled up to a threefold improvement in these data-handling operations, allowing flash storage to be used to its full potential without creating bottlenecks.
Other companies face similar challenges. OpenAI, in developing its Jalapeño chip, focused on minimizing data movement altogether by keeping entire workloads within one connected system. Though the approach differs from Nvidia's, the underlying goal is the same — boosting efficiency through smarter data routing rather than simply adding more processing power.
This shift toward data orchestration opens a new competitive front, where Nvidia will face rival chipmakers and hyperscalers alike. For now, though, Nvidia appears to hold a significant lead in this emerging layer of AI infrastructure.
