In this interview from MWC Barcelona 2026, Steve Dickens, chief executive officer of Hyperframe Research, joins theCUBE's John Furrier to discuss how AI is reshaping telco infrastructure as operators pivot from 5G monetization struggles toward becoming full-fledged AI platform providers. Dickens highlights that most AI use cases showcased at MWC center on cost reduction — RF optimization, power management and observability — rather than new revenue generation. He breaks down the blurring lines between telcos and cloud providers, noting that operators like TELUS and BT are positioning themselves as neoclouds to capture enterprise AI workloads, particularly in sovereignty-driven European markets.
The conversation also explores the broader infrastructure refresh opportunity, where dormant power across thousands of central offices and tower sites could support distributed AI inference at scale. Dickens and Furrier examine why inference is the defining AI trend of 2026, with Hyperframe Research's 544-respondent global survey confirming that a hybrid, multi-cloud, multi-model approach is taking hold. The discussion turns to which companies have made winning bets — from Cisco's return to double-digit growth fueled by AI networking demand to Broadcom's balanced semiconductor-and-software business under Hock Tan's lean operational model. Both push back firmly on the "SaaS apocalypse" narrative, arguing that Wall Street is over-indexing on surface-level signals while missing the deep data moats and network effects that protect incumbents like Salesforce. From the dormant power sitting in carrier facilities to the convergence of cloud, telco and AI platforms, Dickens provides a practical framework for understanding which infrastructure bets will define the next phase of the industry.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
MWC Barcelona 2026. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open the link to automatically sign into the site.
Register for MWC Barcelona 2026
Please fill out the information below. You will receive an email with a verification link confirming your registration. Click the link to automatically sign into the site.
You’re almost there!
We just sent you a verification email. Please click the verification button in the email. Once your email address is verified, you will have full access to all event content for MWC Barcelona 2026.
I want my badge and interests to be visible to all attendees.
Checking this box will display your presense on the attendees list, view your profile and allow other attendees to contact you via 1-1 chat. Read the Privacy Policy. At any time, you can choose to disable this preference.
Select your Interests!
add
Upload your photo
Uploading..
OR
Connect via Twitter
Connect via Linkedin
EDIT PASSWORD
Share
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
MWC Barcelona 2026. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open the link to automatically sign into the site.
Sign in to gain access to MWC Barcelona 2026
Please sign in with LinkedIn to continue to MWC Barcelona 2026. Signing in with LinkedIn ensures a professional environment.
In this interview from MWC Barcelona 2026, Steve Dickens, chief executive officer of Hyperframe Research, joins theCUBE's John Furrier to discuss how AI is reshaping telco infrastructure as operators pivot from 5G monetization struggles toward becoming full-fledged AI platform providers. Dickens highlights that most AI use cases showcased at MWC center on cost reduction — RF optimization, power management and observability — rather than new revenue generation. He breaks down the blurring lines between telcos and cloud providers, noting that operators like TEL...Read more
exploreKeep Exploring
What are the key themes you’re seeing in the telecom industry—particularly regarding 5G monetization, network slicing, and use cases like stadiums and events?add
How has the culture within telecom companies been changing as they struggle with 5G monetization and move toward cloud‑native, hybrid cloud, edge AI/accelerated computing—and in light of early discussion about 6G?add
What proportion of the use cases you're hearing about are focused on network optimization and cost savings versus enabling new, monetizable business services?add
What infrastructure trends and business opportunities are emerging from AI, workflow changes, and recent events—particularly around efficiency, decentralized/hyperconverged edge computing, and using/retrofitting telco facilities to host GPUs?add
How can telcos take advantage of the shift to inference-focused AI infrastructure, and what architectural/networking changes (including the roles of hyperscalers and vendors like NVIDIA and Cisco) will that require?add
Will the rise of AI cause a "SaaS apocalypse" and make established SaaS companies like Salesforce obsolete?add
>> Welcome back, everyone, to theCUBE's live coverage. This is day four live coverage of MWC, formerly Mobile World Congress, 2026. I'm John Furrier, host of theCUBE. Dave Vellante's out on assignment, he'll be back shortly. As we wrap up MWC, we're looking at all the data coming in. It's obviously been a really complex, overwhelming show from a content standpoint. All the boxes are being checked. Enterprise, AI in the telco, AI in the network, the radio networks. Hyperconverged Edge, the report I shipped earlier in the show. Here to break it down and put it back together is Steve Dickens, CEO of HyperFRAME Research, friend of theCUBE. Steve, we're always chatting. We see each other at all the analyst events, but as we pull back and wrap up the show, you've been scouring and doing meetings back-to-back, like all the other analysts that we feature here. What are you seeing at the show? And put some context around it because we cover AI like it's nobody's business with NVIDIA, Broadcom, all the chips, Marvell. You name it, from the bottom of the stack all the way up to the top where the gen AI-native apps are. And so, that's great, but now you have the telcos who control the networks and the operators. You got the infrastructure and you got the carriers. The operators are certainly making a lot of money in North America, but not so much maybe in the sovereign states, but sovereignty's a hot topic. So, so much to unpack. What does your analyst notebook tell you even though you have a phone? It's not really a notebook anymore, but what did you discover? What were some of the things that you surfaced?
Steven Dickens
>> So, hey, John, thanks for having me on the show. We've been talking about doing this for a while, we haven't got round to it, so it's great to be on the show. So, really, just I think three or four key themes that we've been seeing. Obviously, monetization. We're starting to get towards the end of the 5G rollout, 6G's on the horizon, 5G advanced. So, as we get into those later stages, really fascinating conversation with Vodafone yesterday. We picked up on some of that with Intel this morning, basically, around how are these telcos planning to monetize their huge investment, not only to buy the spectrum, but then to build out the infrastructure. So, still struggling with that monetization, network slicing, some of the advanced SLA features. We've seen a lot of stadium and event use cases. Can you offer premium services? Still really hard to get end users to pay, get a stadium to pay for enhanced 5G coverage through network slicing, that's proving to be difficult. So, I think they're still struggling with 5G monetization. And then, obviously, we've got 6G starting to be talked about. Maybe three years away, but people are starting to plan ahead and think-
John Furrier
>> It might be sooner. 6G, I mean, at terrabit speeds, you're talking about 6G, I think the 6G could happen faster if the AI factories comes to the edge, which we've been saying. We'll find out at GTC. I know NVIDIA's going to have some announcements. But I think the key thing I'd like to get your thoughts on is that what's the culture shift luck been like? Because we've been seeing and reporting and you see it as well, that the culture shift from selling connectivity or revenue per circuit, per subscriber, the economics have always been laid out in front of the telco, but they can't ignore what's coming to their doorstep, which is cloud-native, hybrid cloud. We heard IBM talking about Red Hat and automation feeding into telcos. Again, that's a good tell sign because IBM's on the ground got some ground truths, but the bigger picture is this wave of accelerated computing and intelligence fused into telco is hitting them. So, you can't ignore the punch in the face. So, you either-
Steven Dickens
>> 100%....
John Furrier
>> you take it and you run with it. And again, they got the keys to the kingdom. They got the data, they got the network. There's no KV cache in networking, but networking concept of KV cache in the Blackwell, NVIDIA example, or InfiniBand, those interconnects, they have the connections. They could put it all together. They got virtualization. What's your take on that?
Steven Dickens
>> You asked the question around what's been one of the big things. Obviously AI, it's 2026. It's been pervasive across the conversation. Kicking off on Sunday with Justin Hotard, the CEO of Nokia, talking about how they're partnering with the likes of Intel and AMD to bring AI to the edge and to bring it actually into the core network. Then, we spent some time with Intel this morning talking about some of the use cases for AI in the radio access network, being able to deploy tiny language models, hundreds of thousands of parameters, not even billions of line parameters, around how to optimize frequency and wave-forming and beam-forming. And we nerded out this morning, but I mean-
John Furrier
>> RF is an easy area where quantum, AI-
Steven Dickens
>> RF optimization....
John Furrier
>> will really be a big difference because that's connectivity coverage, that has to be there because physical AI is right around the corner. Robotics and self-driving autonomous cars need connectivity. So I mean-
Steven Dickens
>> Yeah. We're going to be pushing more and more inference out to the edge. Then, the other piece is the AI-RAN Alliance here. We've been hearing a lot from our conversations with Vodafone. We just literally came from a session with Oracle. So, I think really AI has come into this vertical and it's all layers on the stack.
John Furrier
>> What's the use cases that they're presenting that you're getting briefed on? Because there's two things we're seeing, and I want to get your thoughts on this. One is improvement to the things in networks like frequency, optimization, AT&T laid out some stuff here in theCUBE. But then, there's also the business side, because one of the things we heard from IBM and others is that, "Well, 5G didn't monetize very well. We're not too sure about how this other stuff, until we see use cases."
I don't see a lot of business conversations around, "Hey, these are compelling services that me," say Ericsson, "could offer." Because AT&T runs on Ericsson. So, they're making a ton of money, but is Ericsson dipping into that pool? Maybe, maybe not. So, the monetization conversation is not so much, how do we monetize the thing? It's how do we use the thing to make money? So, it's kind of like, how much of the use cases, percentage-wise, are optimization, making our stuff run better, to actually compelling new business services or monetizable, enabled, whatever you want to call it? What would you say?
Steven Dickens
>> Well, you're asking exactly the right question. No surprise there. I think the thing for me is I've heard more about that optimization piece. The three cases that I just came from the Intel briefing and the demos this morning around bandwidth and RF optimization, optimization of power, being able to turn cores off based on demand, and then being able to optimize from a OSS from an observability point of view. All of those use cases were cost reduction use cases. I've heard very-
John Furrier
>> Or squeezing efficiency out of it, like power.
Steven Dickens
>> Yeah, squeezing more efficiency. But I mean, I use a simple rubric. Value props need to be make money, save money, not get fired. All I've heard this week has been save money. And this goes back to my point earlier around the 5G monetization, that's been the challenge. How do we get money to be generated out of that?
John Furrier
>> I would add another one to your list, how to get promoted? Because whoever makes the money gets promoted. Again, I look at it different... By the way, the efficiency piece is huge because I don't think that's bad. I mean, that's like just where we're at. So, the question is, how fast can you go? Because one of the common themes we heard on theCUBE this week, and we've been hearing this in the AI and also the crypto world and the infrastructure side is you can change the workflow value chains and not have to change the outcome of what you're trying to do. In other words, there's too many process pieces. That's one. And the other one is this refresh wave coming. They're always talking about, which is AT&T, Verizon, all the operators have central offices. They got towers. Those are facilities. They got cabinets, Steven. So, you put a little baby factory in there and you add it up and you got grid computing. I mean, the amount of dormant power that's out there, we estimated on our tally on our research side that there's about 22,000 GPU potential just laying around power-wise. Enough power to power 22,000 GPUs. That's about a $20 billion a year revenue stream estimate. That's just sitting idle. Now, the caveat is you got to upgrade. So, retrofitting some of these pieces is a huge opportunity. I mean, think about 22,000 GPUs. How many data centers you have to build to do that? Now, they'll still build the data centers, but you've got that distributed power. And if you look at what's happening in Iran, Amazon going down, the case for hyperconverged edge, in my opinion, is so strong because Amazon got taken out in Bahrain, Middle East, because they didn't have decentralized infrastructure. So, distributed infrastructure will happen. I think the war that's going on will point to that. I mean, it's been like a spotty conversation, but given the current events, that wasn't on my bingo card. But I think it points to the fact that if you're Amazon or a hyperscaler or building a data center, if you don't have a distributed architecture, you're screwed.
Steven Dickens
>> Yeah. So, I think outpost, I think moving inference... I think 2026, we saw this from Lenovo at CES launching three dedicated inference-focused servers. I think if you look at what the big mega-trend that I'm going to see for AI, I mean, we've still got optimization, we've still got the training piece happening. There's still work going on there, but I think it's at that inference and it's moving it further out towards-
John Furrier
>> Any inference right now. I mean, that's right now. The demand for inference is high, you would say that?
Steven Dickens
>> It's huge. It's huge. So, we launched some data yesterday. We launched a global survey, 544 respondent global survey. All of that's available on our website.
John Furrier
>> What did it say? What was the results?
Steven Dickens
>> So, basically, it's still that we're still so early in this phase. I think we're further along than we were, but we're pivoting from that early POC to adoption. But I think the point that you're making that was well backed up in our research is this hybrid approach. People are going to have multiple cloud providers, multiple infrastructure providers and multiple models. I don't think it's a winner takes all. Maybe there's some winner takes most at the GPU level all around NVIDIA, but I think you look at Google are going to offer TPU-as-a-service and unbundle that-
John Furrier
>> How does the telcos take advantage of that?
Steven Dickens
>> Well, I think a lot of it-
John Furrier
>> Because they're turning into AI platforms, basically, managed platforms. So, if you take a playbook out of the AI infrastructure way past three years, past two years in particular, all the build out on the AI infrastructure with and what NVIDIA has done and others is enabled the new use cases. So, there is no AI native without NVIDIA and OpenAI running on the cloud hyperscales that they got. So, what is the equivalent... I guess my question for you is, what is the equivalent AI infrastructure metaphor for telco? Are they actually staring at solutions? Are they trying to figure it out? Because if they don't get their infrastructure game up, I'm talking about enabling it to be AI-enabled, towers and the RF opportunities are just as massive. How are they going to back all... The stadiums are easy to talk about because it's dense population, we all know the need there. But is there an build-out mindset on the infra? Do you see that?
Steven Dickens
>> Yeah. So, I mean, a lot of these telcos, you look at somebody like TELUS, they're a managed service provider, arguably neocloud provider and a telco provider, BT. So many of these telco providers are also cloud providers, so I think you're going to see them building out just to pick up those enterprise use cases. And obviously, that's a-
John Furrier
>> I think you're right on. I think they are cloud providers. They got network-
Steven Dickens
>> Those lines are definitely blurring.
John Furrier
>> I mean, hybrid cloud is basically distributed computing, so that's computer science. So, if you're going to run a cloud, say in France or Spain or have a sovereign nation-
Steven Dickens
>> You mentioned a really key point. I was at a Cisco Live EMEA, probably half of the conversations there were sovereignty. I picked that up again. Obviously, we're in Europe this week. A lot of the conversations are sovereignty. So, if you're a local operator in a geo and you've got a cloud component to your business, there's real strong demand for some of those sovereign workloads and being able to... We're seeing that line blend between telco operator and neocloud and MSP. And I think with sovereignty as a mega trend, I think the telco operators that run a cloud business is going to benefit.
John Furrier
>> I've had a lot of non-disclosure meetings with some alpha engineers on both sides, cloud, hyperscale, neoclouds and telco. And these are people making the architectural decisions. And the reason why was is that it's not an incremental game right now. The telcos have always played that incremental game. "5G, 6G, get the towers, do a refresh, blah, blah, blah." It's an architectural shift. And so, I was looking for some signal around, "What is that architectural shift?" And everyone has come back and basically said the following, "It's a war for networking." I go, "What do you mean?" And they said, "Well, look at NVIDIA. NVIDIA's networking numbers are off the charts." Yet I asked Jeetu Patel here and sitting in that chair-
Steven Dickens
>> 10% growth in Cisco's last earnings.
John Furrier
>> I'm like, "Jeetu," I go, "How do you relate with NVIDIA? Because you had Jensen up on stage at a big event in San Francisco?" And so, I wanted him to answer the question because, and he answered it right. He didn't say, "We're not competing with NVIDIA. They're networking within the large-scale system or supercomputer or AI cluster. They're horizontal, they're connecting, like they always do, networks, campuses, et cetera, et cetera."
So, if you apply that networking theme to telcos, what does Cisco do? What do they do? What are the telcos do? Who's going to be the provider? Now there's the edge where the box ends and the RF picks up and then you got the far edge, which is the radio side, I'm oversimplifying it, but you have two networks coming together. So, you got radio frequency and you got wireline. Now, the wireless side, you got multi-protocol access, Wi-Fi, 4G, 5G, 6G, satellite, you got fiber, all these things are feeding into the wireless side, so it's got to come together. And if you put an NVIDIA 1U box factory, which I think they're going to announce at GTC, I mean, come on, you got an AI factory with intelligence pumping tokens out. What does that do in your opinion?
Steven Dickens
>> Well, I think all-
John Furrier
>> Do you agree with that concept?
Steven Dickens
>> Yeah, I mean, absolutely. I mean, I think we're seeing some of these lines, we're just talking about it, some of the lines between cloud and telco blurring. I was on the HPE booth, chatting to them about their new very dense ATX 12000 launch that they made this week. So, you're starting to see some of these foundational technologies. I mean, Cisco reported 10% growth. I think that's probably in their earnings, what was that a week ago? When was the last time we saw double digit growth from a Cisco? So, there's definitely an accelerant as with Gordon, who runs their EMEA business-
John Furrier
>> Well, the demand. They're feeding the AI demand.
Steven Dickens
>> Feeding the demand for sure and they're seeing that as an accelerant. So, I agree with you, I don't see NVIDIA's business as competitive to Cisco, or Juniper for that matter, part of HPE now, obviously. I think it's a rising tide is raising all boats scenario.
John Furrier
>> Yeah. Cisco's a good point. I want to ask you in your expert opinion, because Cisco's a great example of bets that paid off.
Steven Dickens
>> Absolutely.
John Furrier
>> So, we're seeing markers now on earnings where bets are paying off. So, who has made good bets? Because the horses are coming in, right? So-
Steven Dickens
>> So, I think you had Jeetu on the show-
John Furrier
>> He's obviously made great bets....
Steven Dickens
>> he's box office. I spent some time with him in EMEA at the Cisco Live event. He's doing a fantastic job. I think not only on being a fantastic spokesperson for the company, but also-
John Furrier
>> He dresses well.
Steven Dickens
>> Oh, that Tom Ford jacket at Cisco Live, he's operating in a different class than-
John Furrier
>> Jeetu, you're going to get the best-dressed award. You're looking pretty good with that pocket square there.
Steven Dickens
>> If I was giving analyst awards for best-dressed executive, I think Jeetu's going to win that. But all joking aside, John, I think Cisco's got its mojo back, probably for the first time in 20 years.
John Furrier
>> We'll see. I'm not sure about that.
Steven Dickens
>> I think it's foundational around some of the cross-collaboration. He spent some time with us in a closed-door session talking about how he's driving cross-functional collaboration. You look at what they've done with products like AI Canvas, the focus on Silicon One and they're driving at the semiconductor level. We're really seeing a different Cisco and that's, I think start... So, some of the bets that Chuck made on some of that-
John Furrier
>> Who else made good bets? Obviously, NVIDIA makes all the best bets, they're leading the market. Broadcom, did they make the wrong bet? Where are they at?
Steven Dickens
>> I mean, Hock Tan is probably the best financial operator in the CEO class. I think whilst they've taken some knocks on the VMware, I spent some time with those guys. The level of innovation, they launched a new telco version, a VCF9 this week. So, a very radically simplified portfolio from those guys. There were 20,000 SKUs prior to the acquisition, they're now down into single digits-
John Furrier
>> If you look at Broadcom in the past six months, okay? If you look at even year-to-date, they hit over 400, okay? They hit 400 and change. They're dropped now almost 100 points. They're 313 right now.
Steven Dickens
>> Is that just wider market though? I think we're seeing the SaaS apocalypse, we're seeing it come through.
John Furrier
>> There's no SaaS apocalypse. I think that's BS.
Steven Dickens
>> I don't buy that either. I really don't buy that.
John Furrier
>> Yeah, it's BS.
Steven Dickens
>> I mean, we tried to do some work to agentically create a CRM internally the other week, completely failed. Here's the problem. Nobody's going to come after Salesforce.
John Furrier
>> I guess we'll pound the table together because we both agree. My opinion on this is that the Wall Street doesn't get the AI value proposition yet because-
Steven Dickens
>> Well, we saw this with IBM just last week with their 13% intraday drop because of a-
John Furrier
>> They're making the buying opportunities very strong for everybody, but look at-
Steven Dickens
>> IBM's already back up.
John Furrier
>> Jensen said, actually at the Cisco event, he said, "Things aren't spreadsheetable yet." And I put that in quotes because he means no one can model out the scale or any kind of discounted cash flow or any analysis because it's forming because it's happening in real time. It's unfolding as we see it and they're emerging, but the benefits are real. So, the SaaS companies are recognizing that, yes, the SaaS model of downloading an app in the Apple Store might go away to a prompt or, "Hey, load hyperconverged edge," just automatic stuff happening, but the software apps will happen. So, that's not a question of the company's going out of business, that's a discussion of the user interface. So, Salesforce won't go away because they have a moat. They have a huge data moat.
Steven Dickens
>> It's the data. It's the data.
John Furrier
>> So, my analogy was, and this came from a friend who worked at Apple with Steve Jobs and co-founder of Nest, he said iPod was the number one selling product when iPhone was being developed and everybody at Apple was like, "Well, we're making all this money with the iPod." Jobs was hardcore, Steve Jobs said, "We're doing the iPhone." He created the iPod, the phone, the computer, and the app store platform in one device. That product change was an upgrade to the iPod and created more functionality. That wasn't a SaaS apocalypse, that was, "Here's the new value proposition for the market that wants it." And RIM, Blackberry and Nokia had all the same opportunities to do the same thing, they did not. That is not a function of AI or some technology, that's a business model decision. So, SaaS companies that are out there have to decide, do they evolve to the new thing? And yes, some stuff that's low-hanging fruit that's easy to replicate, that has no competitive durability to it, yeah, that's going to get eaten up by an entrepreneur. But replicating Salesforce, that's like saying, "Hey, I want to create a Reddit competitor." All the code is open source. I can just load Reddit called CUBEit-
Steven Dickens
>> It's the network of effect....
John Furrier
>> but how do I get the users? So, there's no Reddit out there because you can't replicate the user base. So, the SaaS thing is just ridiculous. That is a dumb Wall Street mindset. Wall Street does not understand tech. I'm there now, most of the people there just don't get it. But they'll get it because when the money hits the table...
Steven Dickens
>> Yeah. Well, we saw... I mean-
John Furrier
>> That was a rant. Sorry about that.
Steven Dickens
>> And I'll just go and I'll do the same-
John Furrier
>> Yeah, go ahead. Answer the rant....
Steven Dickens
>> because I think we're kindred spirits on this. I think we saw this with IBM last week, a Claude blog about COBOL drives a 13% decrease. I mean, look at where IBM is back up. If you pull the numbers up, the week view, they've probably recovered all of those gains. The last time I looked, they were up and recovered at least half of it. So, I think the market's skittish. As you say, John, I think the market's skittish. People are looking for signal and they're over-indexing on particular parts of the signal. I think for me, it's a case of you and I track those fundamentals. We look at where these companies... We talked about Salesforce, the moat and the connectivity and how they're fundamentally baked into how these companies operate. If you've got a 2,000 person sales force out there chatting to enterprise buyers, you're not going to be ripping out Salesforce anytime soon, you're absolutely not. Whilst that interface may change, and we're going to see Agentforce, and we're going to see that user interface change and evolve over time, absolutely that moat is crucial. So, what I don't think is being factored in is Salesforce from a development point of view are going to increase velocity. They're going to be leveraging AI. Their code velocity's going to increase. Their functions and features that they're able to bring to market will increase. So, we're exactly on the same page, John. I don't buy into this SaaS apocalypse narrative.
John Furrier
>> Let's talk about earnings. I know you do a lot of earnings. Broadcom just had a nice clean beat. Wasn't the blowout quarter. What is your take on them? Is that an inline analysis? Obviously, the semiconductor space is exploding. Hock Tan reset expectations for the Q2 guidance, AI run rate's looking up, and also validates the demand on the AI CapEx. I see that to be a tailwind for Broadcom.
Steven Dickens
>> You look at Broadcom, I was literally just with their team last night, they've announced a new structure under Hock that I think reflects the organization. He's put two SVPs in, one for the hardware business and one for the software business. So, they're now a very radically-different business post-VMware acquisition. 50% of their business is coming from software, 50% is coming from the semiconductor side. I spend a lot of time, as you can imagine, with the mainframe team, knowing my coverage, they're absolutely crushing it. Then, you look at the VMware piece. I think they're post the industry noise now, and they're solidly in execution. And then, you look at the hardware business, as you say, massive tailwinds for that semiconductor business. So, I think Hock runs that business nice and lean.
John Furrier
>> They nailed the semi-side. So, I mean, they're going to be catching up quickly to NVIDIA, AMD as well.
Steven Dickens
>> For sure.
John Furrier
>> What I love about Broadcom, and we've covered this on our side too, is that their seasonality piece is great. When semis are up, VMware's doing good, maybe not growing. VMware goes through the cycle, cyclical nature of-
Steven Dickens
>> They're a very balanced business.
John Furrier
>> They have a very nice cyclical balance. And when it's all working, like when it hit 400... And I think the NVIDIA sucks a lot of oxygen out of the room on Broadcom. Hock doesn't do a lot of press interviews. Charlie's pretty quiet about things. But they've been quietly getting more out there on the press circuit.
Steven Dickens
>> Yeah, I'd describe them as Mag 8 and my addition would be to include Broadcom.
John Furrier
>> All right. Well, Steve, great to see you. Final summary, how would you wrap up the show? What's the headline on your report?
Steven Dickens
>> 2026 and it's all AI.
John Furrier
>> It's all AI on theCUBE here. I'm John Furrier, Dave Vellante. Day four, live coverage, breaking down the analysis, digging into what's going on in telco. This is a convergence of many things enterprise, consumer, telco, cloud, AI all coming in. Very complex show, a lot of moving parts, but it's infrastructure game. Once the infrastructure's built, the results come out. The key is will the telco build out the right AI infrastructure? We'll be following it and we'll bring that to you. Thanks for watching.
John Furrier