-
Cisco said it's putting together AI systems to take the technology from 'science project to business-critical'
-
The vendor is using gear from Intel, Nvidia and Red Hat among others
-
Cisco said that many verticals and enterprises are using existing AI models
Cisco told Silverlinings that it is gearing up to help enterprise customers as their AI pilots go from “science projects to business-critical projects.”
To that end, Cisco has just announced an expanded roadmap of Cisco validated solutions based on proven industry platforms to provide customers with a pathway towards AI. Cisco is using hardware and solution from AMD, Intel, Nvidia, Nutanix and Red Hat, along with others to put together these systems.
"One thing that we’ve observed in our enterprise customers is that very, very few of them are developing and training these large foundational models on their own,” Todd Brannon, senior director of product marketing, compute told us. “Typically we see them grab an existing model and run with it.”
This is different from how some telco AI models are already beginning to evolve.
So customers could use a model from Hugging Face, AWS, Meta or Open AI, among others, drop the data in, tune the infrastructure and take it from there, Brandon said. Obviously, this may cut down on the compute power and power requirements needed for AI models, which, as Silverlinings has already seen can be extreme.
Cisco said that AI is already growing in the retail, manufacturing and healthcare sectors. The company noted that retail is particularly using AI Inference already. Inference is the process of running live data through a trained AI model to make a prediction or solve a task. This can be used in retail to help best position the products to make a sale.
In time, Brannon said the affect of AI will be felt across all industries.