Nvidia's edge is shifting from GPUs to the systems around them

For the first years of the AI boom, Nvidia was the only source for state-of-the-art GPUs, which became highly profitable as the industry scaled. In recent years, hyperscalers like Amazon and Google have started building their own chips, leading investors to question how durable Nvidia’s advantage really is. After growing its market cap 10x between the start of 2023 and mid-2025, Nvidia shares have moved on a more modest trajectory over the past year amid concerns about GPU competition.

A new narrative has taken shape since Nvidia’s earnings on Wednesday: investors are recognizing that the company’s advantage extends well beyond GPUs. As AI compute scales to the gigawatt level, orchestrating it has become an increasingly complex task, and Nvidia has built much of the state-of-the-art hardware needed to handle it — giving the company an advantage in the systems surrounding the GPU even as GPU competition intensifies.

Nvidia is currently rolling out its Vera Rubin architecture, which pairs the Rubin GPU with other units including the Vera CPU, the Groq 3 LPX inference accelerator, and matching racks for storage and networking. According to Nvidia, these are extremely specialized systems that, rather than processing tokens themselves, ensure everything outside the GPU runs as efficiently as possible. The Vera CPU in particular focuses on orchestrating data. “Vera is important because there’s only so much memory that you can put in a single server or any sort of compute platform,” Nvidia’s VP of storage technology, Jason Hardy, told TechCrunch, adding that the company saw “upwards of 3x improvement” in operations where the Vera CPU accelerates access to flash storage without bottlenecking performance.

A similar problem shows up outside Nvidia. In a blog post earlier this month, OpenAI said its Jalapeño chip was designed to minimize data movement and communication delays by keeping an entire workload within one connected system — a different approach to the same underlying goal of gaining efficiency through smarter traffic control rather than more processing cycles. Nvidia will still have to compete with rival chipmakers and hyperscalers at this new layer, but at least in the early stages, it appears to hold a commanding lead.

Why it matters

If the real battleground shifts from raw GPU performance to system-level orchestration, Nvidia’s early lead in that layer could offset the rising competition it faces on chips alone.