How VMware and Broadcom Are Securing the Future of Private AI

The security landscape is dramatically shifting as organizations embrace private AI and cloud environments, creating new vulnerabilities alongside unprecedented opportunities. Umesh Mahajan VP & GM, Application Networking and Security Division, Broadcom pulls back the curtain on this evolving threat landscape where generative AI is democratizing attack capabilities: "suddenly machines can be used and anybody can become an attacker."

Ransomware attacks don't appear from nowhere—they represent the culmination of methodical intrusions where attackers establish footholds through back channels before executing their final devastating blow. As VMware Cloud Foundation and Kubernetes-based private AI deployments become standard, Broadcom's security strategy centers on tight integration, ensuring WeDefend and AVI Load Balancer provide seamless protection without performance compromise.

Modern security demands distributed approaches rather than centralized bottlenecks. Traditional physical firewalls and load balancers can't efficiently secure today's dynamic workloads without introducing latency and complexity. Meanwhile, enterprise IT teams face mounting pressures—infrastructure grows exponentially while staffing remains static. This reality drives Broadcom's focus on automation, simplification and intelligent assistance.

Perhaps most exciting is how Broadcom harnesses AI itself as a security enabler, developing LLM-powered assistants that transform how teams troubleshoot problems and identify security gaps. These tools interpret natural language queries about runtime states, blocked traffic, and performance issues, while proactively identifying unused security features that could strengthen protection. For compliance-focused organizations, they've mapped capabilities to frameworks like NIST and developed air-gapped solutions for highly regulated environments.

Ready to strengthen your security posture for the AI era? Explore how these innovations can protect your private cloud and AI environments while simplifying management, meeting compliance requirements, and keeping performance at its peak. Subscribe to hear more conversations with security leaders navigating today's most challenging threats.

Chapters

• 0:00 Introduction to Application Security
• 1:29 Security Challenges in the AI Era
• 3:17 Performance and Scale Requirements
• 5:16 Simplifying Security with Intelligent Assists
• 7:33 Compliance Solutions and Future Innovations

Transcript

Speaker 1: 0:01

And we are here at VMware Explorer Blockbuster Day. Umesh, good to see you again. Good to see you again, evan, too. Congratulations, what a week it's going to be. Before we dive in, maybe introduce yourself and your team within Broadcom.

Speaker 2: 0:17

I'm Umesh Mahajan. I'm the General Manager for Application Networking and Security. Essentially, we have two products over here WeDefend, which is our security product, and the RV Load Balancer. So that's what we focus on.

Speaker 1: 0:31

You make it sound so simple, but there's so much complexity to decode here. Let's talk the big picture. More workloads are moving to private AI, private cloud environments, and what are the implications for security in this new world that's rapidly?

Speaker 2: 0:49

approaching. One thing is security is top of mind because the attacks are only increasing every day, they're not decreasing, and with AI, and more so with Gen AI, you shudder at the thought how fast these attacks will come in and how complex they will get in, because now suddenly machines can be used and anybody can become an attacker. So security is top of mind because compromises are happening and all the way till ransomware, there's malware attacks, but ransomware attacks are the most serious and they show, you know, the gaps in security at various levels. But the ransomware attack suddenly just doesn't start from nowhere. You know the attackers come in from some back channel and then slowly make their way up and then the ransomware attack happens. This has been happening and it's going to like just accelerate with AI rate with AI.

Speaker 2: 1:51

So I think, as we look at private cloud and VCF, the ask post-acquisition was that this division really focused on providing the best security and the load balancing solution for VCF, private cloud, and as private AI takes off and is beginning to happen, it's Kubernetes-based. How do we continue to provide that same level of focus for security and load balancing? In load balancing there is a security part also, right? Uh, there is the waft part and, uh, post quantum cryptography. We are doing so.

Speaker 2: 2:19

We take both the products. We defend this 100 security. But part of our V is also security focused and those parts are becoming important suddenly because of all the attacks WAF and the other layers. So we are very much focused on these aspects and we are seeing a lot of value to the customers because we are plug and play. We just fit in nicely and architecturally as VCF moves forward 9.2, 9.1, they're completely integrated and we fit in compared to some external security or load-basing products. So that makes us a natural fit and we are able to take advantage of some of the building blocks to provide higher level of security Fantastic, and application performance is critical, as you know.

Speaker 1: 3:08

How do you enhance performance and reliability and security without sacrificing anything in that balance?

Speaker 2: 3:17

So I think performance and scale is something that our customers have demanded for two reasons. We are a software-defined product, so we run a controller at the top. So the more we scale and the more performance we can push through, it's of higher value to the customer. It drives down the cost. And the other thing is the hardware is moving. The server hardware is moving right x86, et cetera, et cetera. New versions come out and the number of cores, so you can pack in a lot more applications of Kubernetes, workloads on the same server. So we need higher scale and higher performance over there.

Speaker 2: 3:57

And the other thing is the customer teams are not growing Like our customers. The staff is not growing, so they want to go to the same controller and be able to manage a much larger estate, to be able to automate from that one place and manage it with small teams. So scale and throughput is becoming super important. As we go down new features and private AI and whatever you call it. These workloads are going to drive a lot of throughput and you have to secure them in a distributed fashion because you can't go to a physical firewall or a physical load balancer, because it's very complex to set it up. And and then you would add latency and time and you know all those kind of problems. So you have to be able to solve this issue in a distributed fashion in a very optimal way, because performance and throughput and latency, along with providing the security, are super important as this journey to private AI and agentic AI is happening.

Speaker 1: 5:06

Wow, it's quite a balancing act, and so security and networking, as you describe, are coming together. And how do you simplify this for enterprise customers, Because they also want an easy button. You know they want simplicity as well.

Speaker 2: 5:17

Yeah. So a couple of things over there. One is we are simplifying the whole deployment journey Because customers, because the limited staff sometimes are, you know their adoption cycle takes more time than it should, Because, hey, there's something you need to do. They agree, they see the value of lateral security zero trust but takes them time. How do we simplify all the steps? So we've come up with prescriptive steps with full visibility. We'll ask a few questions, Tell me this mapping. After that we'll auto-generate the rules and we'll step by step make them do it so that we shave off like multitude months and you know a lot of complexity that they have to get it done. So that's one big focus both for security and load balancing we are doing. The other is um, which is in tech preview. Is this whole simplification is with llms.

Speaker 2: 6:13

You know chat boxes are the vogue, so we have intelligent assists coming out, for we defend the DFW firewall and also the RV side. This is what will help the customer. They can ask any question. It's not just documentation, that's the easy part, but runtime state what is happening here, what is blocking this application, why this traffic is not getting through, why the performance over here is lower. So this triage as they decide, and then we'll know they've turned on these three features, but they haven't turned on the fourth and fifth feature. While they're asking questions, we can tell them hey, not fully secure, are you sure you don't want to turn this on because you have the license for it.

Speaker 2: 6:57

So we are gonna do all the. We'll roll out just a guide. Help the customer quickly and they don't need to worry like, oh, how will I figure out if there's a problem? They can deploy with confidence. We'll simplify that. And when they run in, you know something fails or mistakes are made configuration when that happens, they'll have full confidence that the product itself will help me figure out what is the right approach, how do I rectify it? What do I do? So I don't need to worry. I don't need to have tons of experts on my team to make progress over there. So give them that built-in support so that they can move forward with confidence.

Speaker 1: 7:38

Wow, very impressive. Of course, the flip side of security is compliance, and rules are tightening everywhere and you work with some very big distributed global enterprises and healthcare and telecom places where compliance and security are paramount, but you don't want to slow development innovation. How do you manage through that challenge?

Speaker 2: 8:01

So one is like you know how do we meet, help meet some of the compliance requirements right, and sometimes we find that the visibility and, like IDS, IPS, the IDS part, helps them meet some compliance requirements. At least they should be able to detect what is going wrong. So we are encouraging our customers to turn on these elements so that they can meet compliance needs. Or, hey, even distribute firewall If they've deployed it and customers see, okay, you have protected these workloads properly, then you can meet the compliance need.

Speaker 2: 8:38

So we are looking through those standards and telling the customers if you use our product to do this, this, this, you will meet X, Y, Z parts of the compliance. And that helps them in the regulatory areas. And sometimes we are also doing air gap products because they can't go to the cloud for some feeds. So how do you have them, those signature feeds, available on-prem so that they can meet the regulatory compliance need? So we are mapping nicely there's a NIST framework, et cetera what elements of it our product addresses, so that we don't do everything in the security in the world, but what we do, we want to help them meet these compliance. That's very fundamental.

Speaker 1: 9:24

So, as we head into the rest of Explore and beyond, what are you most excited about over the next weeks and months?

Speaker 2: 9:30

So I think you know we have several new innovations which are rolling out. So I met several customers here. We even had a customer cab on Friday and Saturday. The feed yes, saturday and Sunday, sorry. The feedback was all positive on our roadmap, our innovations and the journey we are taking. They said this is exactly what we need. I'm very excited about the work we are doing and how we will have WeDefend and Avi be the best firewall security solution and load balancing solution for VCF and as that journey continues, we will remain in lockstep and provide our customers tremendous value in these areas. As we move forward and all our innovations and intelligent assists, and as this journey evolves to more AI-driven workloads, we are quite confident we can address them with the work we've done already with Kubernetes and security and load balancing and as it evolves, we are ready to take advantage of it.

Speaker 1: 10:36

Well important work. Congratulations on all the success and onwards and upwards, Thank you.

Speaker 2: 10:42

Thank you so much.