Cisco and Nvidia push enterprise AI beyond the hyperscaler stranglehold
Source: SiliconANGLE
The AI infrastructure market is bifurcating. Hyperscalers no longer have exclusive access to the rack-scale hardware and software stacks needed to run serious language models and foundation models at enterprise scale. Cisco's networking layer plus Nvidia's accelerators and software are creating a path for mid-market companies to run their own AI workloads without renting compute from AWS, Google, or Microsoft. This threatens the hyperscalers' margin capture on inference and fine-tuning workflows. The shift redistributes AI economics: enterprises keep their data and model improvements in-house, while Cisco and Nvidia move from component suppliers to full-stack orchestrators competing directly with cloud providers.