Day 1 Insights from KubeCon + CloudNativeCon NA 2025
Hosted by Savannah Peterson and Rob Strechay of SiliconANGLE Media Inc. and theCUBE Research, the first day of KubeCon + CloudNativeCon NA 2025 begins with engaging discussions on the evolving role of Kubernetes in AI infrastructure. With thousands in attendance, conversations focus on artificial intelligence, particularly the potential of Kubernetes in driving AI inference solutions for diverse businesses.
Savannah Peterson, Principal Analyst & Host at SiliconANGLE Media Inc., and Rob Strechay, Director/Principal Analyst & Host at theCUBE Research, facilitate dynamic interactions with guests Youp and Ned. They explore core themes of AI inference, observability, and Kubernetes' role in a cloud-native environment. Peterson emphasizes the significance of a robust technology foundation in the context of AI-driven systems.
Throughout the day, the necessity of integrating observability and security in AI deployments emerges as a critical takeaway, reflecting the complexity and dynamic nature of these systems. Discussions highlight the shift back towards hardware considerations and the intriguing potential of WebAssembly in future technology roadmaps, indicating a blend of both established and emerging technologies.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
KubeCon + CloudNativeCon NA 2025. 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 KubeCon + CloudNativeCon NA 2025
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 KubeCon + CloudNativeCon NA 2025.
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
KubeCon + CloudNativeCon NA 2025. 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 KubeCon + CloudNativeCon NA 2025
Please sign in with LinkedIn to continue to KubeCon + CloudNativeCon NA 2025. Signing in with LinkedIn ensures a professional environment.
Are you sure you want to remove access rights for this user?
Details
Manage Access
email address
Community Invitation
theCUBE Insights Day 1 with Ned Bellavance, Ned in the Cloud & Joep Piscaer, TLA Tech
Day 1 Insights from KubeCon + CloudNativeCon NA 2025
Hosted by Savannah Peterson and Rob Strechay of SiliconANGLE Media Inc. and theCUBE Research, the first day of KubeCon + CloudNativeCon NA 2025 begins with engaging discussions on the evolving role of Kubernetes in AI infrastructure. With thousands in attendance, conversations focus on artificial intelligence, particularly the potential of Kubernetes in driving AI inference solutions for diverse businesses.
Savannah Peterson, Principal Analyst & Host at SiliconANGLE Media Inc., and Rob Strechay, Director/Principal Analyst & Host at theCUBE Research, facilitate dynamic interactions with guests Youp and Ned. They explore core themes of AI inference, observability, and Kubernetes' role in a cloud-native environment. Peterson emphasizes the significance of a robust technology foundation in the context of AI-driven systems.
Throughout the day, the necessity of integrating observability and security in AI deployments emerges as a critical takeaway, reflecting the complexity and dynamic nature of these systems. Discussions highlight the shift back towards hardware considerations and the intriguing potential of WebAssembly in future technology roadmaps, indicating a blend of both established and emerging technologies.
theCUBE Insights Day 1 with Ned Bellavance, Ned in the Cloud & Joep Piscaer, TLA Tech
Savannah Peterson
Principal Analyst & HostSiliconANGLE Media, Inc.
HOST
Ned Bellavance
Independent Consultant & Technical EducatorNed in the Cloud
In this KubeCon + CloudNativeCon North America Insights segment from Atlanta, theCUBE’s Savannah Peterson and Rob Strechay wrap up day one with analysts Ned Bellavance and Joep Piscaer, unpacking how AI is reshaping the Kubernetes conversation. They highlight why inference is becoming the real AI battleground, how Kubernetes is emerging as the common layer for running those workloads, and why observability and security are now core pillars of the emerging AI conformance spec – including the need to monitor prompts and responses, not just CPU and latency. <...Read more
exploreKeep Exploring
What is the primary focus discussed in relation to artificial intelligence and Kubernetes?add
What is the significance of observability in scaling systems, particularly in the context of AI?add
What are the implications of the shift in technology towards integrating AI with hardware in collaborative partnerships?add
What was the impact of the AI boom on the progress and perception of WebAssembly (Wasm)?add
What observations have been made about conversations in the open source community compared to other discussions?add
theCUBE Insights Day 1 with Ned Bellavance, Ned in the Cloud & Joep Piscaer, TLA Tech
search
Savannah Peterson
>> Good evening Cube fam, and welcome back to Atlanta, Georgia. We're here at the end of day one of our three days of coverage at KubeCon. My name is Savannah Peterson. Delighted to be joined by three really fantastic analysts. More importantly, two of which are my friends. I feel like we just get to hang out. Ned, you're about to become my friend. Ned, welcome to the show. Great to have you.>> Thank you, future friends.
Savannah Peterson
>> Yes, future friends.>> Good to hear it. Yes.
Savannah Peterson
>> We've also got Youp and Rob, fan favorites. And, we all got the purple memo today, which I secretly love. It's conference colors. So it's fantastic. It's been a really exciting day one. There's allegedly over 9,000 people here. Conference center is buzzing. Ned, I'm going to open it up with you. What stood out with you since KubeCon kicked off the most?>> Obviously, the focus is artificial intelligence, AI and how Kubernetes can accelerate and help in different dimensions. The one they chose to lead with was inference, which I thought was interesting. Typically, I would think of something consumer-facing, or something in the training models, but they were like, "No, we want to talk about inference and how important it is." And I think it has something to do with the potential market for inference versus training. There's only a few companies that are going to do full training, but the number of companies that are going to need solid inference, that is huge.
Savannah Peterson
>> It's everybody.>> It's everybody.
Savannah Peterson
>> Yeah, if you're using AI, inference is what makes it real to the end consumer, to your point, or to the developer, or whatever that might be.>> Yeah, so building out a solid tech foundation for you to roll your inference out on. And also, they were talking about making it agnostic to which platform you're using as long as it's Kubernetes. That was an interesting push as well. And, it seems like the project is very much at the 1.0 stage, but there's already some validated solutions.
Savannah Peterson
>> Youp, I see you smiling as he's talking. What's going through that beautiful brain of yours?>> So I agree with Ned. I think it's interesting that they're going for that angle, simply because that is where the money is. That is where basically everything is going to be running and Kubernetes is the place to be. And I like how it resonates and vibes throughout the whole show floor. Everyone's talking about AI, but we're at a stage where we're no longer saying, "We do AI." But we're at a stage where people say, "Yeah, we do AI, but we do observability." For instance. And so, I like how observability is now a thing again. It always has peaks and valleys and it meanders, but now observability is back and I like how we keep on going through these motions.
Savannah Peterson
>> Well, it's really mission-critical. I mean, observability always matters, but as you're ramping up and scaling, especially with AI, we're talking about a lot of things going on within a system and a stack. And if you can't see what's going on, it could be pretty wild. Rob, has your experience been similar? You've been all over the place, you've gotten to talk to me, you've got some briefings.
Rob Strechay
>> Yeah, I mean I think-
Savannah Peterson
>> You've had a day....
Rob Strechay
>> I've had a day.
Savannah Peterson
>> You've had a day.
Rob Strechay
>> I've had a day. I agree on the inference thing. I think that's where the money is and that's why it sense. The Neo Clouds are where people are going to train their models and do all of that stuff, and they're here too, but there's a handful of them. You don't need to go market to them as much. But, to your point on when we were talking, just last night we were at Infrastructure's Code Conference or IACConf Connect. And, some of the conversations afterwards, even the last question we got there was around, "Hey, how do I do observability basically for prompts?" And stuff like that. I think, the complexity of AI and so many moving parts has everybody coming to the table, because everybody's freaked out. I mean, I was having another discussion just even about what are the identities these agents are going to use and how are you going to manage and audit all the identities of an agent? So the security factors, we've had a couple of those discussions today.
Savannah Peterson
>> Yes, we have.
Rob Strechay
>> I think, it's one of these things that I think it's passed the infrastructure Kubernetes and gotten to the AI infrastructure in a lot of that.
Savannah Peterson
>> The whole AI conformance spec that they were pushing, two of the big pillars were observability and security. And I think that speaks to your point. We have these golden signals we normally observe for key metrics we're looking out for, "What's my CPU utilization? What's my response times on things?" Now there's a new metric we have to watch, which is, the prompt and also the response. Does the user like that response? Is it a valid response? If I asked you for, I don't know, directions on how to get from point A to point B and instead you give me directions on how to get to point C. I don't see that in my normal metrics, but that's a wrong answer, and that's a bad signal to be giving. If I do that too many times, you're going to leave my service. So we now have to include the responses as part of the whole observability stack. That's going to be tough and people need to start instrumenting for that.
Rob Strechay
>> And John, there was a analyst lunch and Jonathan came in, and basically it was funny, he was reviewing what he had talked about at the keynote where we were all standing in line trying to get in. But, anyway, besides the security snafus that have been going on here today and-
Savannah Peterson
>> Security's a hot topic these days, Rob, whether it's RRL or your data....
Rob Strechay
>> It definitely had a firewall up and we're constraining the flow of traffic into the place today. When he was talking about it, it was very interesting, because I think it was a lot of, exactly where you were going with that, the observability and the security, but it was also, "Hey, we don't quite understand how we're going to get there." And part of it he brought up was VLLM and LLMD, which are not quite yet under CNCF. I have a feeling they're going to land either with CNCF or Linux Foundation somewhere. But, I mean, just because of the history and who owns now, I think that it will get steered in that direction, which is great, because then you start to get a center of gravity around this, but there's so many different places you have to go, I think right now. And, from an open source perspective, that confuses people, and it makes their lives a lot more difficult about how did they get to observability, how did they get to security on that stuff?
Savannah Peterson
>> And don't forget, the world is circular. Everything comes back. And I think next year, or maybe even in Amsterdam for KubeCon, we're going to be talking about hardware again. We've gone away from hardware, everything was cloud, but we're now in a place where we have to consider 400 gig networking, because the models need stuff like that.
Savannah Peterson
>> I am so glad you just said that, because we were just talking about that on my last segment with Google Cloud. And two things, so this hardware is back. Hardware had gone into the whisper nut there for a little bit, as if all this stuff isn't running on hardware. Like, "Yeah, cool." The cloud is hardware, spoiler. But the thing that was also really interesting about that, and talking to the GKE team specifically is they're trying to be hardware-agnostic. And what's really interesting about that is it's a shift and that's a paradigm like, "Okay, we've got to make sure that we're ready for whatever people are going to plug and play within their partnerships."
And it's an extension of the culture of collaboration that we've been saying on the software side to power all this stuff with AI in particular. But I love it, because I am an OG hardware nerd and it's been a little lonely not getting to talk about it all the time. And, it's, oh nice right now, because all of a sudden, especially with the inference conversation, the inference is going to happen on hardware that people are touching, and doctors are doctoring with, and things are happening, but it's going to be Kubernetes and all that we do is going to be built into all this stuff in a way that's intuitive and not so in your face or necessarily shouting that it's there. But in a way that's optimizing the performance of that solution, which is very, very exciting. I'm curious if there's any conversations that you were expecting to have that you haven't really had a lot of so far. You probably go with you, because you've been to oodles of coupons.
Savannah Peterson
>> So the discussion I think hasn't happened yet is, what's next after Kubernetes? In the before times, before Kubernetes, we would have this discussion every few years, right, "What's after bare metal? What's after virtualization?" I'm missing that conversation. And I think that is because Kubernetes is now so ubiquitous. Everybody's using it in the data center, in the cloud, on-prem, even at the edge, it's everywhere that I don't think we're ready for that discussion yet.
Savannah Peterson
>> I think that's a really good point. We had Kelsey Hightower, obviously one of the grandfathers of Kubernetes. Grandfather feels odd. He's not old enough to be a-
Rob Strechay
>> Godfather....
Savannah Peterson
>> Godfather, yeah. Sorry, Kelsey. Hopefully not watching. I appreciated our segment.
Rob Strechay
>> He's right there.
Savannah Peterson
>> But, what was fun and appropriate about our conversation is the vision of Kubernetes when it came out was to be a 20-year technology, to your point, not to be 5, 10, 15 years, but to actually be something that saw us through a couple different technological hype curves and implementation. And so, I love that you brought that up, because I think instead we're at this calm piece right now, where everyone's like, "Cool, so this is what we're building with. Now, let's go build that whole future." Ned, what about you?>> Well, it's funny you should mention that we're in this lull. Because I felt like before the AI boom hit, there was a lot of talk about Wasm and web assembly.
Savannah Peterson
>> I was going to bring that up actually, yeah.>> And it was getting some strong headwinds. You saw it making progress.
Savannah Peterson
>> It was a whole moment. Oh, yeah.
Rob Strechay
>> And I thought that was the next big technology. That was the thing that was not going to replace Kubernetes, but it was the next big thing people were going to be hyped about. And, potentially could replace some Kubernetes workloads in a new form. And then, AI dropped like a bomb and took all the wind out of Wasm sales. That's not to say that people haven't been working on it. There has been a lot of progress made. I was talking to somebody today who's putting together an open source project around Wasm and replacing container run-times with WebAssembly instead for MCP servers.
Savannah Peterson
>> Mm-hmm.>> So, when your MCP server needs to reach out and use a tool, instead of it reaching out to use a container, it could use a Wasm module instead. It's smaller, it's more lightweight, and probably more secure. I was like, "Ooh, that's a great idea." But yeah, there really was that period of time where I thought it was going to be all WebAssembly all the time. And, we moved away from that. Maybe things are getting to the point now where AI is starting to cool down and we're ready for the WebAssembly thing.
Savannah Peterson
>> I think it's going to be really interesting. Yeah. I know you and I feel very similarly about this.
Rob Strechay
>> We do.
Savannah Peterson
>> Well, to your point, I know there was Wasm Day at KubeCon last time, I believe, at least in London, and then potentially even in Salt Lake, if I'm recalling correctly. I haven't seen anything about it here. Not even a whisper. I haven't even said the acronym all day on the desk, and we talked about-
Rob Strechay
>> Well, you were having some discussions with them. I thought we talked about it yesterday, but maybe I'm-
Savannah Peterson
>> We might've talked about the modules, and the security, and how tiny it is, and how it really goes in the sandbox, and you can be so prescriptive around what a Wasm process can do down to Syscalls, which is amazing. It also requires a lot of backend support to get it working properly. It's very fiddly at first. And so, I think, they were trying to shake out that fiddliness. And then, yeah, all the investment shifted to AI and I won't say it died in the vine, it's still there. It just needs another injection of life, I think....
Savannah Peterson
>> Yeah. There's not a lot of fertilizer right now.
Savannah Peterson
>> Yes, it's been scooped up by something else.>> It'll have its moment....
Rob Strechay
>> Yes, it'll come back around. To your point, I think it depends on what you want to put into it, and I think that was always my thing with Wasm was is it going to really replace certain runtimes or not? And I think that was... Now, there may be reasons to do that, because the ephemeral nature of it, and how you can bring it to different places, and as these AI-agentic applications get built out, it may see another push in there as that.>> It's the same with serverless. The higher you go in the stack, the more prescriptive it gets.
Rob Strechay
>> Yeah.
Rob Strechay
>> And so, the harder it gets to actually run something in there. And so, that's the difficulty of serverless too. We haven't heard much about serverless too.
Savannah Peterson
>> Haven't heard anything about it, at least speaking from here. Yeah.>> Because containers apparently are good enough for so many use cases, and that's what I love about Kubernetes. It's so-
Rob Strechay
>> It's also cost-effective too. I mean, when you look at serverless, I think serverless was originally going to be cheaper than everything else and blah, blah, blah. And funny enough, serverless runs on a server. You still have to... Every time you spin it up. I know.
Savannah Peterson
>> Is there hardware involved?
Rob Strechay
>> There's hardware. There might be. Yeah. But, when you look at it, there's the startup penalties. "Well, if I'm going to have that startup penalty, why am I tearing it down? Because then, I have to go through that all over." So, even some of the database companies that have rebuilt their architectures on serverless, like Databricks in particular, some of the nodes that are in there are still long-lived containers, and not just the serverless, so it's not all serverless because the architecture doesn't work that way and it can't.
Savannah Peterson
>> Yeah. Well, no, technology really replaces the previous one. It augments it for certain use cases. Just because Kubernetes came along, it didn't mean VMs went away. What do you think most of those Kubernetes clusters are using? It's all virtual machines.
Rob Strechay
>> Right.>> But we move up a layer in the stack and we add some more abstraction. There needs to be a benefit to that. And I think that's where we're trying to figure out what is the true benefit for WebAssembly versus what we're using today to put all this additional effort into actually making it work.
Rob Strechay
>> Yeah, agreed.
Savannah Peterson
>> I like that. One thing that I've noticed, I mean, you and I have been on the road a lot. We have been show to show to show, back to back to back. And everywhere else we've been talking, it's been a lot about agentic. And, it's interesting to me here, and this is one of the reasons I love the open source community is they don't talk about hypothetical. They talk about what's really happening and what we're really building today, and it's been refreshing. Of course, there's some GKE announcements about agentic, plenty of things in the news. But the conversations have been much more around how we're optimizing workloads, how we're doing inference, to your point. It's been around this, "How are we making sure that this foundation that we're making with this first ramp up of AI..." Not first ramp up, but you know what I mean, more ubiquitous adoption of AI, that this is going to be strong enough to handle when we add the layer of agentic that creates both chaos, complexity, and a whole bunch of other stuff? And it's nice. Because I just love that we can really talk about things that are real and we're not listening to, I say this with love as a consultant, consultants talking about the 60,000 agents that we've deployed, you're forcing your employees to use them and it's really a chatbot. Is that a agentic? I don't know. I mean, sure, I guess. But, not in a way that applies. So I think, for me, it's been really refreshing to hear about the fine-tuning, to hear about the power optimization, to hear about how we're really making sure that this is the right stable table for us to then go dance on with all our friends and make all the cool solutions. Last question for you gentlemen. What are you most excited to dive into more here on day two? Some of the conversations, there's dinners, there's events. Rob, I'll start with you.
Rob Strechay
>> Agentic.
Savannah Peterson
>> Yeah. How did I know I could count on you today? Something smart like that.
Rob Strechay
>> Yeah. No, I mean, I think again, I think there's a lot still going on. And, I brought this up, how we talk about agent being the wrong name.
Savannah Peterson
>> Yes. That's why we love the genies.
Rob Strechay
>> Genies.
Savannah Peterson
>> We love genies.
Rob Strechay
>> Well, I called it buddyless. And, you look at it and go... I think people haven't figured out the use cases for it yet, per se. Not everybody. And it's not widely, and the ROI still isn't there for a lot of it.
Savannah Peterson
>> No one knows what they're doing.
Rob Strechay
>> Yeah. So we're early days with that. But I think to your point about hardening the infrastructure underneath it, I think there's a lot of work that needs to be done. Again, observability, security. I think, identity is a massive one. I was talking to CyberArk about that earlier today in a briefing, and that was literally what we were talking about, because I'm sitting there going, I can wrap around easy a chatbot. I can't wrap around easy a... Something that needs credentials to go... I'm in accounting. I can go and build a dashboard. But then, Sue from marketing comes in and asks the same prompt of the same agent and doesn't get the same thing back, because she doesn't have the credentials. How do you stop that, where you have to build it that it understands and has a tokenized identity down to the thing? I think, that will probably be... When we look at Amsterdam, I think that'll be one of the number one things that people are talking about, because something bad's going to happen, and it's going to cause people to go, "Wow."
Savannah Peterson
>> Just gave me chills.
Rob Strechay
>> I hate to be a pessimist on that, but most-
Savannah Peterson
>> No, I mean, you're not pessimist, you're accurate, you're an analyst....
Rob Strechay
>> Something bad has to happen to really shift the power over to that.
Savannah Peterson
>> Well, and I think just to double down on what you just said, in terms of it being in Europe, it tends to be a community much more regulation-savvy, much more aware of the risks. In America, we're like, "Well, we'll just see what happens." Until the whole world shuts down, all the flights are grounded, and healthcare stops. Well, that's just another day in the United States of America these days. But I do think you're right about that. I think that's a really good point. Youp, what about you. What are you excited for?... >> And so, I'm going to follow your train of thought there. Put it in a different perspective, because I'm an ops guy at heart. I like hardware, I like keeping systems operational. So, what I like about this show is that is a continuing theme. Sometimes you have to look for it. Sometimes it's not really there, but you have to go and dig it out, and so I'm interested in figuring out the operational aspects of all of this. How do people actually keep all of this stuff running and how to do that well? That's super interesting to me. That's going to be my next thing.
Savannah Peterson
>> Yeah, it is super interesting, and there's so many different manifestations of that on the show floor with all the different projects, all the different scalers. I mean, you can have different forms of that conversation, which is cool, which is why it's so fun to learn from this community in particular. Ned, what about you?
Savannah Peterson
>> I'm going to borrow from that a little bit. The conversations we were having last night and today all had to do with human in the loop. AI is important, but we can't trust it. We need to have a human in the loop, and there's so many humans here. I want to talk to them and find out how they're inserting themselves into the loop to keep their AI safe. That's been my conversations today. And I want to have more of those conversations tomorrow.
Savannah Peterson
>> I think that sounds great, Ned. I know we are the humans in the loop right here.
Savannah Peterson
>> How about yourself?
Savannah Peterson
>> Ooh, good question. I think that I love how grounded this conversation is, but I would really love some real world examples of how this has manifested. And I've pushed some of our guests on it today like I always do. But, I know they're out there and I know there are some cool things that are happening, and so I'm hoping that tonight and tomorrow people come up and start talking to me and saying, "Hey, look at this thing." Or, "Look at how we applied this." Or, "Look at this X-ray technology we can now optimize and people can get the results way faster." Because that's the stuff that really gets me super excited. I mean, the hardware is the core of that, but the outcome is what I really care about and I am waiting for us. I feel like we're at the top of the roller coaster before it starts to go down, and it's really exciting where I'm like, "Come on, come on, come on. Show people that this works, show it that it works, show it that it works, so that we can go whee and go down and do the whole thing." Which is where we'll be next and less worried about the details. But, yeah, if you've got those stories and you're listening, please bring them to us here on theCUBE, Ned, Youp, and Rob, I just love doing this with you. Ned, we're now friends, so we've crossed that chasm.>> Excellent.
Savannah Peterson
>> This has been absolutely fantastic. And thank all of you for tuning in to day one of our three days of coverage here at KubeCon. Quick shout-out to all the veterans out there. Thank you all for your service. We appreciate you and we see you. My name is Savannah Peterson, signing off from Atlanta, Georgia. You're watching theCUBE, the leading source for enterprise tech news.
Rob Strechay