Rami Rahim

CEO

Juniper Networks

AI-driven networking takes center stage as data demands soar

AI-driven network automation is reshaping the way enterprises scale their infrastructure to meet the demands of artificial intelligence.As AI workloads surge, networks must evolve with smarter, autonomous solutions to handle rising data demands. Technologies such as liquid cooling, co-packaged optics and high-performance fabrics are now essential for AI-driven data centers. To stay competitive, organizations must enhance efficiency, scalability and resilience by adopting AI for networking to streamline operations and networks for AI to support massive GPU clusters, according to Rami Rahim (pictured), chief executive officer of Juniper Networks Inc

play_circle_outline AI for networks and networks for AI opportunities

play_circle_outline Importance of Mist AI in simplifying network operations

play_circle_outline Juniper's history and expertise in silicon development

play_circle_outline Energy efficiency and power optimization in networking

play_circle_outline Impact of silicon in technology industry and innovation

Info
Transcript

Rami Rahim, Juniper

Rami Rahim

CEO Juniper Networks

In this engaging interview, Rami Rahim, CEO of Juniper Networks, shares his insights on the cutting-edge developments within the networking industry at MWC25 Barcelona. Conducted by theCUBE analysts Dave Vellante and Bob Laliberte, the discussion delves into how Juniper is harnessing AI to revolutionize network operations and cater to the ever-increasing demands of artificial intelligence. Rahim elaborates on Juniper's strategic initiatives, addressing the transformations in customer data centers and the shift towards AI-centric infrastructures.

Key ta... Read more

explore Keep Exploring

What are the two big opportunities in the context of artificial intelligence and networks? add

What sets our solution apart in the industry today and allows operators more freedom to focus on advancing their strategies and generating new revenue streams? add

What are the three critical ingredients of an AI-native network according to Juniper? add

What is Juniper's approach to achieving performance while also focusing on power efficiency in their networking technology? add

What gives a competitive advantage in IT and chip development, particularly in terms of artificial intelligence innovation? add

bolt Powered by CUBE AI

Rami Rahim, Juniper

Dave Vellante

>> Hi everybody. Welcome back to Fira Barcelona. My name is Dave Vellante. I'm here with Bob Laliberte. Really excited to have Rami Rahim back on theCUBE, CEO of Juniper Networks. Rami, thanks for coming on.

Rami Rahim

>> Great to be here.

Dave Vellante

>> Good to see you. Last year, a year ago, we were on this set and of course we were talking about the acquisition. We don't really want to get into the acquisition. A lot of that's been said out there. But what can you tell us about the state? A lot of information is out there in the internet, but give us your quick take.

Rami Rahim

>> Look, I'll say that we remain super excited about the proposition of combining with HPE and becoming a really powerful player in the industry, especially in networking. We're working through the process, including the DOJ challenge. That being said, there's so much amazing things happening at Juniper right now, I'd love to focus this interview on that.

Dave Vellante

>> Yeah, good. Well, let's do that. What's got you excited these days? Where do you want to start?

Rami Rahim

>> Juniper started in the age of the internet, and I remember way back then we were really pushing the edge, the limits of technology, silicon development software, scale out the software development and so forth. Fast forward to today what's happening with artificial intelligence, I feel like once again we're having to push the very limits of what's possible with technology. Things that were kind of nice to have, like liquid cooling, co-packaged optics and so on, are now becoming necessary to have to keep up with the pace of change and the capacity requirements of AI.

Dave Vellante

>> So how are you seeing your customers deal with that change? Are they refactoring their data centers? Obviously the hyperscalers are all over this. The on-prem customers that we talk to, they all want to build AI centers of excellence. They're rethinking their air-cooled data centers. Because they really haven't invested a ton. They've been de-investing in data centers for a while. What are you seeing?

Rami Rahim

>> I think there are two big opportunities. The first one is AI for networks. And that really is an opportunity that transcends customers. It applies to every customer segment, cloud, SB, enterprise. And this is all about leveraging artificial intelligence to make network operations far easier, to basically make the heroes of network operations really successful for their organizations. But to also delight the end user with an experience that's always awesome, irrespective of what they're using the network for. That's the first big opportunity. The second opportunity is networks for AI. This is now basically making artificial intelligence applications of any type possible to begin with. Here, of course, you need to have this high performance fabric connecting GPUs, tens, hundreds, thousands, hundreds of thousands, soon to be millions connected together. You need a network with the right set of attributes to keep up with that capacity.

Dave Vellante

>> So how is the role of the network's operation person changing? I think about storage, like somebody managing LUNs, that changed overnight. Well, it felt like overnight with the cloud. How is the network operator changing? You got AI for networking, makes their lives easier, but then you got networks for AI. That's a whole new game. How does that change?

Rami Rahim

>> Well, starting with AI for networks, any big CTO, the CTO of a big company, CIO, is going to be typically struggling just keeping up with maintaining the network to provide a great experience for their end users. Like keeping the lights on typically is an arduous task in and of itself. Providing artificial intelligence and moving much of that operations to robots, basically software that's doing that work that would otherwise have to be done by humans, is what this opportunity is all about. Our solution, which is driven by Mist AI, is truly unique in the industry today in giving operators that freedom to focus on much more consequential, important things, like advancing their strategies, keeping the disruptors out, generating new revenue streams. That's what they should be focusing on, not just the day-to-day firefighting that's necessary to delight their end users. On the other side of the equation, networks for AI, there it's just about keeping up with the capacity. I mean, what's happening right now is unbelievable. The pace of investment, tens of billions, hundreds of billions of dollars going into learning. In time all of that learning has to translate to inference and value generation. That's necessarily going to happen closer to where the data is at the edge or even in the customer premises. And that's a whole new skill set that we're helping our customers achieve.

Dave Vellante

>> Now, Mist AI has always been unique in the industry. I want to ask you a question about when you see the race to AGI, all the LLMs, they're leapfrogging each other, everybody's catching up. It's like the NFL, it's like the copycat league. How are you able to maintain your lead with something like Mist AI? Or has the competition, are they closing in? I'm sure you're going to say they're not. But why is that?

Rami Rahim

>> I think there are three really critical ingredients of an AI-native network. I know a lot of people like to use that term, but at Juniper we like to be very specific about what defines an AI-native network. The first is you must have access to the right data. And by data, I'm not just looking at whether the network is up, whether my network elements are working, I need to understand in real time whether my customers are actually happy, whether the experience of using the network is a good experience. That is not something that many people can do. I mean, Juniper I think is somewhat unique in that area. Second, it's about having a proven cloud that can scale from the smallest to the largest of customers, as we have done, again, with incredible wins around the world. And the third is the accurate response. Because some of our peers like to talk a lot about observability, and observability is important, but it's only half of the equation. You must translate observability into insights, and insights then translate into actions. And those actions have to be tangible, reduction in trouble tickets, reduction in time to deployment of new services and networks. And then we're doing that in spades for customers around the world.

Dave Vellante

>> How do you do that? How do you turn those that insight into action? And what role will agents play?

Rami Rahim

>> So AI comes down to data and learning and then translating learning into actions. We have been doing this longer than anybody else. We have been collecting the data now for 10 years, if you include the years that Mist was an independent company. And then since we acquired Mist and expanded that AI architecture across all aspects of our solution. So we have been collecting that data and learning from real life deployments longer than anybody else. That gives us a unique advantage.

Dave Vellante

>> So Jassy's law applies here, no compression algorithm for experience. It seems like it's being challenged in LLMs, but not in your space, certainly not in silicon.

Rami Rahim

>> Yeah.

Bob Laliberte

>> Yeah.

Dave Vellante

>> Go ahead, Bob. I know you want to jump in.

Bob Laliberte

>> Yeah, no. The process has been really what we've talked about is the mystification of Juniper. And when we talk about the differentiation, it's really been the extensibility of the Mist AI engine, that you started with wireless, but it's ... When you talk about what's differentiated, they've got a single AI engine that can go across wireless, wired, data center, WAN, and the innovation just keeps on coming. And a lot of the things that you had talked about, it was really interesting to hear about the AI for networking and networks for AI. And the networks for AI is actually two components. There's the front end and the back end component of it. And you could argue that certainly on that back end environment, AI for networks is critically important for that because you're talking about billions of dollars as an investment, needing to keep it always optimized, self-healing, self-optimized network. And the only way to do that in these environments is through AI. And I'm really glad that you brought up that it's been a 10 year process.

Rami Rahim

>> Yeah.

Bob Laliberte

>> Because so many times when things get exciting, there's a lot of washing, right? AI washing. "Hey, look, we've now got AI ops." Okay, great, when did you start?

Rami Rahim

>> Correct.

Bob Laliberte

>> Right. And so when you talk, you can really tell the difference from an organization that has a mature solution that's able to say, "We've gone back 10 years." A lot of the processes that you've developed over that time really shows that you've got a mature solution that's delivering real value, like you said. And the initial value is always that we've been able to reduce the number of tickets. The real value is with the time saved from firefighting those tickets, they've been able to work on strategic initiatives.

Rami Rahim

>> Yes.

Bob Laliberte

>> And we've seen that the research shows 93% of organizations came out and said, "The network is more important to achieving our business goals." So they recognize that. You're giving them the tools to drive those business outcomes through the operations.

Rami Rahim

>> Yes, yes. Look, it's hard to follow that up because I can't agree more with everything you just said. That being said, you're absolutely right. The strategy for us has been to make Mist AI way beyond what it initially was, which was Wi-Fi, to now include every single aspect of the networking stack. We've expanded it to include wired switching, network access control, SD-WAN, security in the cloud, security on premises, data center. And at this event we're now talking about Mist AI basically simplifying the operations of WAN networks. The other thing that you said that's really important, when you look at networking inside of the AI data center, the most precious resource are those extremely expensive GPUs. And any congestion in the network is evil because it results in an inefficiency, wasted cycles in those GPUs. Nobody wants that. And so we have built into our automation capabilities for the data center the ability to detect congestion and to proactively alleviate it before it's starts to reduce the utilization of those precious GPU resources. That has resulted in some of the wins that we're achieving now in that space.

Bob Laliberte

>> Absolutely.

Dave Vellante

>> Look, I know there's like a backlash on DEI and ESG, but you guys have never been about virtual signaling. And frankly, most people in the data center business understand the importance of having energy-efficient equipment and software that actually can help fine-tune the system. So we just put out our forecast on the future of the data center. It blew me away when I saw the numbers. The data center's been relatively flat. There's been some share shifts over the years. And then all of a sudden, 2023, 2024, it's a spike up and it appears to be headed on a trillion dollar trajectory. I'm talking all in, power, cooling, everything, networking, storage, compute. And it's growing at a ten-year CAGR of around 15%.

Rami Rahim

>> Yes.

Dave Vellante

>> Which is amazing. But the one big risk to our scenario is energy and not being able to get enough of. Everywhere I go, it's like, "Yeah, well we wanted to bring AI on-prem or we want to build other data centers, we just can't get energy in." So what are your thoughts on that as a blocker? How will the industry deal with it?

Rami Rahim

>> So not that long ago the key metric that our customers cared about was performance, basically bits per second. Today, that's ancient history. It's watts per bits per second. It's how you achieve performance from a power efficiency standpoint. Now we've always at Juniper been good at silicon development. We continue to invest in this area. We continue to develop and release some incredible silicon technology that achieves the performance gains that our customers need to keep up with their requirements. But increasingly, our innovation is going in the architectures to achieve even more efficient networking. And we've added now software layers that has the intelligence to understand how the products we're developing in the network are being utilized, and in near real time managing power, turning off engines that are not being used, for example, in order to just save power. And that today is sort of a nice to have, but I think in the very near future it's going to become an absolutely must have. And we're prepared.

Dave Vellante

>> You mentioned silicon a couple of times. When did you start your silicon journey?

Rami Rahim

>> Well, me personally, when I joined Juniper as the youngest employee and engineer on the team, I was a silicon developer. Actually initially I was a silicon verification engineer, then I became a silicon developer. Then I ran a bunch of silicon projects in the company. So at heart, I'm still a silicon engineer. It's great to see the transition over the years. In the early days of Juniper, silicon was front and center in the industry. Then we went through a period where people were confused. It was like software is eating the world and silicon's sort of no longer-

Dave Vellante

>> Don't invest in silicon, right?

Rami Rahim

>> For us silicon guys, that was a difficult period.

Dave Vellante

>> Yeah, I bet.

Rami Rahim

>> Now silicon is once again the front and center. I mean, this is where a ton of the innovation that's happening in artificial intelligence is just staggering. Honestly, you wake up every day and you don't even know what's coming your way in terms of some new big innovation. And it's exciting time to be in IT. It's exciting time to be a chip developer and in technology.

Dave Vellante

>> And it gives you a competitive advantage obviously. Maybe you could explain why.

Rami Rahim

>> So at Juniper, we leverage, depending on the use case, either custom or merchant silicon. In some use cases, let's say data center top of rack switching, we've got great partners, like Broadcom for example, that provide us with wonderful technology that we can leverage that's for our customers exactly what they want. In some use cases, take areas where you need a high degree of flexibility. Our Trio chipset for the MX product line, a product that I actually had a hand in developing back when I was an engineer many years ago, is the most flexible network processor in the planet, that can keep up with new changes and use cases, protocols and so forth. So the investment protection that a customer gets in leveraging Trio for the MX is basically endless. But we've also developed a line of silicon called Express for our PTX product lines that optimizes for power efficiency and performance. This is where our large cloud provider customers, large telcos that are building converged cores, are really turning to as a great solution for their use cases.

Dave Vellante

>> And your process nodes, do you require super advanced manufacturing, like an iPhone?

Rami Rahim

>> Of course.

Dave Vellante

>> You do.

Rami Rahim

>> We're always on the cutting edge of process node technology, of course.

Dave Vellante

>> How important is it for you as a silicon designer to have a US-based advanced manufacturing capability?

Rami Rahim

>> I think it's really important for the US. I think having a healthy ecosystem of manufacturing, basically fabrication options for silicon, is very good. So we would fully support and we would look to leverage any sort of fabrication facilities and capacity in the US. That's going to happen over time. So at this point in time, we're still in a wait and see mode, but I'm quite hopeful and optimistic that that can happen in time.

Dave Vellante

>> I am as well. But so you would agree it needs to be an onshore. I guess it'd be fine in Europe too. But as Americans, at least part American, you'd like to see it in the United States. How important is it that that's a US-based company? Does that matter as much?

Rami Rahim

>> I think it does. I mean for US national interests, having US companies, not just design, but also have the ability to manufacture chips, which are such an important part of the overall technology ecosystem and the economy of any given country, is very important. So as an American, I think this is a very good thing to have.

Dave Vellante

>> So not just TSMC having plants in the US, I'm saying a US-based ... Whether it's a joint venture, which I think it should be perhaps, but US domiciled HQ. You would agree?

Rami Rahim

>> So TSMC having plants in the US I think is great and it's a wonderful step forward. But to your point, I think it's also a very good thing to have a US-based company such as Intel have fabrication facilities available to many different network chip or any chip design for that matter -

Dave Vellante

>> I've laid out my plan for this to happen. I'll share it with you. I've also laid out a plan to keep Intel as an independent brand, as a designer. I hope that happens as well. I don't know how you feel about that as a silicon guy, but maybe you don't want to comment, which is cool. But you have really a vested interest in that.

Rami Rahim

>> Yeah.

Dave Vellante

>> And I think it's good for the world actually if we can do it.

Rami Rahim

>> I cannot agree more. Silicon will always be a key part of where we invest, where we innovate, and where we differentiate. Networking is inherently a distributed problem. There are some aspects of it that can be centralized and moved to the cloud, as we have demonstrated better than anybody else with our Mist solution. But there are some aspects of networking that will inherently always be highly distributed, and you must have purpose-built silicon for that function. And this is where we will continue to invest in our silicon technology.

Dave Vellante

>> Well, and you've seen it. A number of companies have got ... Apple has an advantage, clearly. Amazon, its Annapurna acquisition. Juniper is another great example, a Tesla. I mean, it's silicon is in a way eating the world, right? I mean it's the underpinning-

Rami Rahim

>> It's back in fashion.

Dave Vellante

>> It is so back in fashion.

Rami Rahim

>> Yeah.

Dave Vellante

>> All right, we'll give you the last word, Rami. How do you want to sort of end this segment? What do you want our audience to know about the future of Juniper, its impact on service providers, and the future of AI?

Rami Rahim

>> Well, I will reiterate for a CEO and technology and for a technologist such as myself, it is an absolutely exciting time to be in this industry. Networking is back at the center of the action, both in AI for networks, where we're leveraging AI to simplify network operations and delight the end user. And also it's become absolutely essential to connecting the hundreds of thousands, if not soon to be millions of GPUs in single clusters to power these large LLMs and these unbelievable applications that we're seeing being born around us. So let's see what the future holds. But I'm super optimistic.

Dave Vellante

>> It's interesting you say that, Charlie Kawwas was here last year, same time as you roughly, and he was talking about the future of a million GPU clusters. We've certainly seen hundreds of thousands now. And it's going to happen.

Rami Rahim

>> When Charlie says something, I listen carefully. He's a smart guy.

Dave Vellante

>> Indeed, indeed. Well, Rami, thank you so much for your time. .

Rami Rahim

>> My pleasure.

Bob Laliberte

>> Thank you, Rami.

Dave Vellante

>> And thank you for watching. Bob Laliberte, Dave Vellante, we're here at MWC 2025. This is day two. You're watching theCUBE. We'll be right back to wrap up day two. Keep it right there.