For much of the AI boom, the assumption has been straightforward: as companies adopt artificial intelligence, more computing workloads will move to the public cloud.
Cisco Systems Inc. (CSCO) is now making a different bet.
The networking giant used its fiscal fourth quarter earnings call to argue that the next phase of enterprise AI will not be defined by cloud migration alone, but by companies bringing more AI infrastructure into their own data centers.
The shift, Cisco argues, is being driven by economics as much as technology. As enterprises move beyond experimenting with AI and begin deploying it at scale, they are looking more closely at where AI workloads should run to balance cost, performance and security.
On-Premise AI Infrastructure Is Becoming a Strategic Enterprise Choice
Cisco made its position clear during the earnings call.
“We believe on-premise AI infrastructure will become an important option for enterprise customers as they look to optimize both the business value and cost of AI,” Chairman and CEO Chuck Robbins said.
The statement marks a notable departure from the long-held narrative that enterprise AI would increasingly reside inside hyperscale cloud platforms. Instead, Cisco sees businesses adopting a more flexible approach, deploying workloads wherever they make the most operational and financial sense.
That strategy, the company believes, will require significant investments in enterprise infrastructure rather than cloud capacity alone.
Enterprise AI Deployments Are Expanding Beyond the Public Cloud
Cisco said companies deploying AI increasingly need infrastructure capable of supporting AI applications close to where their data is generated and stored.
As Robbins explained, “Enterprises need GPU clusters on premise and at the edge with low-latency, high-bandwidth networking and built-in security, observability and automation, all of which Cisco Systems can provide in a co-designed, vertically integrated stack.”
For businesses handling sensitive information or applications requiring fast response times, keeping AI workloads on premises can reduce delays, improve control over data and potentially lower operating costs. Those requirements, in turn, create demand for networking equipment capable of connecting increasingly complex AI infrastructure.
Cisco Sees Opportunity Regardless of Where AI Runs
Rather than framing public cloud and private infrastructure as competing models, Cisco believes enterprises will adopt a combination of deployment options.
“Regardless of how or where customers choose to deploy AI — whether in the public cloud, through neo or sovereign clouds, on premise, or at the edge — we believe Cisco Systems will benefit because of the unmatched depth and breadth of our portfolio and our expertise in each scenario,” Robbins said.
Cisco’s outlook reflects a broader evolution in enterprise AI adoption. The first wave of investment centered on cloud providers building massive AI infrastructure. The next may be shaped by enterprises building AI capabilities of their own, creating fresh opportunities for networking vendors as corporate data centers once again become a focal point of technology spending.
