US Markets
Home›US Markets›Sectors›Nvidia builds out AI infrastructure beyond GPUs, inclu…
Nvidia builds out AI infrastructure beyond GPUs, including CPUs and networking
The article says Nvidia’s networking portfolio is its fastest-growing business area after its 2020 acquisition of Mellanox.
Nvidia is positioning itself as more than a graphics processing unit company as demand expands across the broader AI stack, according to an analysis syndicated by Yahoo Finance. The piece credits CEO and co-founder Jensen Huang’s strategy, including Nvidia’s early move to create the CUDA software platform to make its chips programmable for other tasks.
It also points to Nvidia’s role in seeding CUDA into universities and research labs, saying foundational AI code was written on CUDA for Nvidia GPUs. The article frames this history as a contributor to Nvidia’s competitive “moat” in AI model training.
On infrastructure beyond compute, the analysis highlights Nvidia’s 2020 acquisition of networking company Mellanox, describing its networking portfolio as the fastest-growing part of the business and a key step in Nvidia’s shift toward a complete AI infrastructure provider.
The piece further argues that Nvidia prepared for a market shift toward inference and agentic AI by developing ARM-based central processing units, noting that it cites changing hardware needs in AI data centers, including a proposed move from an 8-to-1 GPU-to-CPU ratio toward 1-to-1 as workloads evolve.