Cisco Live: AI takes center stage

LAS VEGAS – Cisco put AI front and center at its Live customer conclave this week, touting new networking, management and security products, along with partnerships and investments it expects will drive enterprise AI deployments.

“AI is moving at an unprecedented pace, especially when you compare it to, say, some of the technologies that we talked about over the years. How long did we talk about 5G before we deployed it?” said Cisco CEO Chuck Robbins in his keynote address. “Generative AI is coming on very quickly, and it’s been deployed across our portfolio, and you’re probably trying to figure out use cases, if you are haven’t already, and how you can use it.”

[ Find your next must-attend event with our guide to network and data center events ]

Ultimately, today’s hyper-extended enterprises will be much easier to manage when it’s possible to look all the way down the stack, all the way into the infrastructure and to the network, to understand what is happening and leverage that data to predict and prevent outages and other problems, Robbins said. Think of the AI evolution as like the cloud transition “on steroids,” Robbins said.

Cisco Nexus HyperFabric AI clusters

Cisco’s AI barrage includes setting up networked AI clusters and using AI to manage networks.  

On the networking front, Cisco added a turnkey AI package called the Cisco Nexus HyperFabric AI cluster, which will be trialed in Q4. The suite will let customers deploy an integrated package of networking in the form of a Cisco 6000 series switch for spine and leaf implementation supporting 400G and 800G Ethernet fabrics, GPUs – in a partnership with Nvidia that includes its BlueField-3 data processing unit (DPU) and SuperNIC – AI pod-building reference designs, and the Vast Data Platform for unified storage, database and a data-driven function engine built for AI.

“HyperFabric is an automated, intelligent AI-native solution, designed to bring simplicity and AI to the on-premise enterprise data center,” said Jonathan Davidson, executive vice president and general manager of Cisco Networking. “The package simplifies the design, deployment, management of networking, compute and storage to build full-stack AI wherever enterprise data happens to reside.”

AI-native Digital Experience Assurance

On the management side, Cisco rolled out Digital Experience Assurance, which extends its ThousandEyes network intelligence platform across cloud environments – Amazon Web Services, for now – and now lets customers pinpoint problems and failures across on-premise resources they own and control, and resources they don’t own, such as those running on a cloud service. Cisco says there are tens of thousands of ThousandEyes agents spread across the internet and enterprise networks, and the platform is powered by more than 650 billion daily measurements globally.

The company also extended its AI-powered cloud insights program. Once customer workloads enter a cloud provider’s network, the platform can show how the components and services connect from a network perspective, how the application load balancers connect to the network load balancers, and how that ultimately forms a target group across multiple AWS EC2 instances, said Joe Vaccaro, vice president and general manager of Cisco ThousandEyes.

AI-driven intelligence in the ThousandEyes package can correlate patterns and unique combinations to triangulate the source of a problem and surface which incidents require attention and which do not, Vaccaro said. “ThousandEyes Event Detection is already providing this critical capability, reducing to mere minutes what could ordinarily take hours and multiple engineers to achieve.”

The company also extended ThousandEyes Endpoint Experience to gather details from Meraki Wi-Fi and LAN telemetry and device information, enabling customers to gain deeper insight into local network issues affecting users, Vaccaro said.

Hypershield support for AMD Pensando DPUs and Intel IPUs

Cisco added support for AMD Pensando DPUs to its new AI-based HyperShield, a self-upgrading security fabric that’s designed to protect distributed applications, devices and data. Pensando DPUs include intelligent, programmable software to support software-defined cloud, compute, networking, storage, and security services. 

Cisco said the DPUs would be available inside Cisco Unified Computing System (UCS) servers and from other leading server vendors by the end of 2024. Cisco plans to support Intel’s programmable infrastructure processing units (IPU) as well. 

Hypershield is comprised of AI-based software, virtual machines, and other technology that will ultimately be baked into core networking components, such as switches, routers or servers. It promises to let organizations autonomously segment their networks when threats are a problem, gain rapid exploit protection without having to patch or revamp firewalls, and automatically upgrade software without interrupting computing resources.

Cisco Security Cloud Control

A new AI-native management architecture, Security Cloud Control, is also on tap. It’s designed to configure, manage and monitor the entire Cisco Security Cloud, beginning with Cisco’s network security solutions.

“Cisco Security Cloud Control will become the single unified management interface for the entire Security Cloud, which will dramatically simplify the security admin and IT experience. We’re targeting initial availability by October with support for our network security products, with more to be added in the months ahead,” said Jeetu Patel, executive vice president and general manager of Cisco Security & Collaboration. The idea is to let customers easily configure, manage and monitor the entire Cisco cloud, Patel added.

“Going beyond AI assistants, Security Cloud Control delivers an AI-native approach to proactively surface actionable insights and automate resolution across hybrid environments,” Patel said.

Within the Security Cloud, Cisco added the 1200 Secure Firewall 1200 Series, which combines advanced security features and SD-WAN connectivity in a single appliance, Patel said.

In addition, a new version of firewall software, version 7.6 of Firewall Threat Defense (FTD), is available for all of Cisco’s physical and virtual firewalls. FTD 7.6 utilizes AI to prevent zero-day threats and extends application control to over 70 generative AI apps to secure sensitive information. It also streamlines branch network rollouts with pre-built SD-WAN and firewall templates as well as supporting zero-touch provisioning, Patel said.

“We believe that we need security for AI, and we need AI for security,” Robbins said. “First of all, you need security to protect the data that you are using to do custom training, as an example, in an AI model that you might be running into private enterprise. You need to have security to make sure that you don’t have threat actors actually infiltrating those models… You need to make sure that the questions that are being posed of these models aren’t questions that are against your culture, or against what you care about. So you need security for AI, and we’re going to continue to work on delivering that.”

AI training and certification

Products aside, Cisco announced plans to bolster AI career-related initiatives. First, Cisco announced the initial stage of its AI partner specialization training that will provide business partners with the knowledge needed to optimize infrastructure for AI workloads using Cisco technology. The training will include AI basics, a taxonomy of AI solutions, and an overview of Cisco’s AI strategy, governance and use cases for Cisco platforms.

Cisco also added a new AI certification in designing AI architecture. The vendor-agnostic certification will let employers, employees and job seekers learn the skills needed to design modern AI/ML compute and networks, according to Cisco.

$1 billion AI investment fund

Cisco is launching $1 billion global AI investment fund to help bolster the startup ecosystem and expand the development of secure and reliable AI solutions. As part of the new AI fund, Cisco announced investments in Cohere, Mistral AI and Scale AI. Over the last few years, Cisco has acquired or invested in more than 20 AI-focused vendors.

As Cisco’s AI strategy unfolds, multi-product solutions and partnerships will play a role in Cisco’s success or failure, experts said.

For example, the Nexus HyperFabric AI cluster has a ton of merit for enterprises, and it is competitively differentiated from other options on the market, said Vijay Bhagavath, research vice president with IDC. “The idea that enterprise customers who want to ramp up AI don’t have to go out and buy all the components to get going will be of value.”

In terms of its partner ecosystem, Cisco can afford to buy or support the companies it needs to build out its AI strategy and help the overall marketplace develop, Bhagavath said. “It has a massive go-to-market edge with AI, but it is early, and how all of this plays out will be interesting to watch.”

Others think Cisco should lay out more of a coherent AI roadmap rather than deliver piece-parts.

“Cisco needs to be very deliberate in the way they attack the AI market and avoid the piecemeal approach,” said Jim Frey, principal analyst of networking for the Enterprise Strategy Group. “Sometimes they look like they are behind some of the other vendors doing AI, such as Microsoft for example, but the reality is they are being very methodical.”

That said, it’s very early in the AI game, and Cisco isn’t alone in developing an overarching plan or roadmap for AI. Many other vendors are doing the same, Frey said.