Cloud Native Maturity Signals Who Is Ready for AI
Cloud native developers are now more than 21 million strong. New research suggests they are also leading the way on AI. The gap with everyone else is hard to ignore.
SlashData’s Liam Bollmann-Dodd and CNCF’s Bob Killen examine that link with Alan Shimel at KubeCon + CloudNativeCon in Salt Lake City. Bollmann-Dodd is Principal Market Research Consultant at SlashData. Killen is Director of Technical Program Management at CNCF.
Two Reports, One Community
Bollmann-Dodd authored both reports CNCF released at the show. The State of Cloud Native Development Q3 2026 estimates the population at 21.2 million. That is the first time it has crossed 20 million.
The second is the CNCF Technology Radar. It captures how users perceive CNCF projects, which doesn’t always match their graduated, incubating or sandbox status.
The AI Readiness Gap
About 75% of cloud native developers use AI, compared with 55% of other developers. Bollmann-Dodd sees cloud native as unlocking the ability to integrate and manage AI in production.
Some non-cloud native developers still reach agentic AI through managed services or APIs. Others may work in cloud native organizations without realizing it.
Killen agrees. Platform engineering often hides the underlying stack, so many people use cloud native without knowing it.
Alan pushes back on the idea that regulated industries lag. In his experience, finance and government often embrace new technology early. Bollmann-Dodd notes that backend developers in finance are right at the cutting edge.
Inside the Technology Radar
This edition covered observability, infrastructure as code and storage. Projects land in adopt, assess or trial based on community recommendations.
Prometheus and OpenTelemetry earned adopt positions in observability. OpenTofu and OpenFeature led infrastructure as code, while Carina led storage.
Killen points out that some Chinese-centric storage projects scored lower than expected. Bollmann-Dodd says the China sample was unusually small this wave. Future surveys will make sure Chinese cloud native developers are fully represented.
Both reports are available at cncf.io.
Transcript
Hey everyone, this is a very hoarse Alan Shimel here at KubeCon. You know what though? I'm hoarse because we've done maybe 40 of these in three days.
It's been a lot of learning, a lot of great information. I hope you've enjoyed it. We got two more to go.
I got these two gentlemen and one more after that, so stay tuned. Let me introduce you to my two guests, and I've already asked them to do me the favor of carrying me on this. So to my immediate left, I've got Bob.
Killin. And Bob, your last name is? Killin.
Killin. And to my far left, I've got Liam. Liam?
Bolman-Doss. Say it again. Liam Bollmann-Dodd.
Liam Bollmann-Dodd. Liam has been on with me before, so he kind of knows the story. Bob, I don't know if we've met.
I don't think so. Excellent. So Liam and Bob are here with me today around two reports that the CNCF has released at the show this year.
Mm-hmm. And I'm going to ask each of them to kind of give us the lowdown on the report. And not to worry, we'll give you the address or URL where you can go get the report for yourself.
Liam, if it's okay, I'm going to ask you to go first. Yeah. Which report are you reporting on?
So I was actually the author of both the reports. So the- Oh, you did both? I did both the reports, yeah.
We don't really need Bob, do we? I'm only kidding. I'm kidding.
So go ahead. So I wrote both the reports. 2 million, so above 20 million for the first time.
We also look into some trends around cloud adoption and technology adoption. And then we did the kind of separate, like deep dissections about like how does being cloud native impact whether you adopt AI, adopt agentic introduction. So that was the first report.
And the second report is the End User Tech Radar, where we picked three kind of CNCF project areas with observability, IaC, and- Storage ... storage. And they basically asked about some projects that CNCF have that are graduated, sandbox, incubated, and got direct user perception of how they think they are, whether you should adopt them, assess it, or try it when it comes to your own projects.
Got it. The difference on those is like, since the products generally have like the sandbox incubating graduated, that doesn't necessarily match with the user perception, and this study gets the user perception of those projects. Got it.
So I actually have used your report, not this year's yet, but I usually do a little recap of that report to have some really good salient figures, because it's a great insight into the community. I'm curious, how the hell are they running agentic AI without cloud native? So, I think there's two parts to it.
I think the first part is that we are in/data, and we're asking developers directly about what you're doing or what you're interacting with, what you're using in your day-to-day work. And it's very possible that someone can be in a team that has its cloud native, but their little corner of it isn't cloud native. So there's obviously a limitation we have, but we talk about it quite openly.
But the second part is that there's often we think a more like a managed... Managed agentic is not quite the right word, but more managed services, maybe API integrations, or it's built into their cloud provider or some sort of access that way. So they're getting access to agentic AI without necessarily having to do the infrastructure themselves.
Got it. Yeah. So the other interesting thing that we saw in the report is the sort of alignment of cloud native, people that consider themselves cloud native developers, and the actual adoption of AI versus the non-cloud native developers.
Whereas cloud native developers have 75% usage, or 75% use AI- Yeah ... versus 55% of non-cloud native developers. What do you attribute that to?
So from my researcher side, there is clearly some barrier or some requirement that cloud native unlocks that allows you to use AI, manage AI, integrate your services, have agentic in production. There is some sort of barrier that cloud native is. And I think Bob can probably speak more technically what that looks like.
And then when you have non-cloud native, what's more interesting, there's even less enthusiasm around coding tools or chat bots, just a generally lower thing, which to me then also reflects what I call maturity issues where some organizations are just not nearly as mature. And then also the last issue, which is more managed or closed industry, so finance, healthcare, government. They're just a bit behind because everything has to go through so many layers of- Compliance and- ...
compliance and associations. I'm going to disagree with you. Okay, interesting.
You know, I've been in technology 35 years- Yeah ... maybe more. And I used to think that way too, that the finance- Yeah ...
the highly regulated industries were always laggards. Uh-huh. They weren't cutting edge.
But I saw it with cloud native firsthand. I saw it with DevOps. Mm-hmm.
Yeah. I've seen it with so many different technology waves. And this goes for government, too.
Uh-huh. You would think the government's like... They embrace these edge tech-- And I don't mean on the edge.
They're the bleeding edge- Yeah ... or new technologies, better, because they're looking for better ways to do what they need to do in this highly regulated environment. Mm-hmm.
And I think this is true with agentic AI. I think finance has embraced- Mm-hmm ... agentics.
I think finance companies who have not yet embraced cloud native- Mm-hmm ... I mean, at this point, I don't want to burst anybody's bubble- No ... but cloud native is not exactly cutting edge.
Yeah. It's been around for 10 years. Yeah.
Yeah. We started this book, right? So let's think about it.
If you're an organization that hasn't adopted cloud native yet- Yeah ... what are the chances that you're going out there and playing with AI? None.
Probably, you've got your stuff sitting in some closet in the back of the room, right? You haven't done cloud. Everything you do is waterfall.
Yeah. You know what I mean? Mm-hmm.
You are the very definition of a laggard. Yeah. Yeah.
So should we be surprised that laggards are laggard? No. Yeah.
That's my two cents for what it's worth. I also think there's the concept of the-- So agreeing with you, among backend developers in our survey, those in finance are always at the cutting edge, like right at the cutting edge. I think it's like the other developers in these orgs but may not be exposed to it in the same way, or they're not seeing it.
So like if you're- Yeah ... I'm not sure, like let's say a web front end for a banking firm, you may not really know if you're cloud native or not because it's not part of your job. Right.
So therefore, you're not seeing it. So I think there's also the complexity of as a developer focus, like research, you often miss what the wider org is doing as well. Yeah.
This actually backs up a lot of the stuff that we found, too, where especially the teams that have adopted more platform engineering and offering catered environments to their developers, we have a large amount of people that are using cloud native but don't realize they're cloud native. Yeah. And they're just exposed to that tiny little fraction of it.
Absolutely. Now, if you told me you were a mainframe developer, right? Because a lot of financial institutions- Yeah ...
right, are mainframe. But we cover the mainframe DevOps space. Yeah.
We're even seeing AI usage- Yeah ... on mainframe. Yeah.
Right? Especially we're seeing, in our survey, there are regions of the world that are much more mainframe and on-prem heavy, mainframe especially. And one of them is like China has a really big mainframe thing.
But what we're seeing is because of the focus China now has on cloud native AI development, AI integration, you have the devs that are like, they only use on-prem mainframe, but they are already doing agentic in production because there is the interest and focus. And these guys, I think often people would often say, oh, mainframe and on-prem are like legacy technologies. And I understand that when you have cloud and all these kind of things.
But I think also there are people who are really, really good at maximizing the value of these things. And therefore, they're not laggards because they're using on-prem mainframe. They are just the cutting edge of an area that most people don't understand.
Yeah. Look, they say AI may be the best thing for COBOL. Yeah.
Yeah. Right? Because you can write COBOL for us.
That was a big problem. Yeah. There's actually a product that recently entered the sandbox on CNCF called Hive, where kind of the point of it is to be this agentic system to help support all these products that have one developer and other weird antiquated, I wouldn't say antiquated languages, but like- I don't know.
Less Qs. Less Qs. Definitely use language.
That's fine. Yeah. Yep.
And that's kind of the point of it, is to be able to do the support and maintenance for those things. Absolutely. Yeah.
Hey, let's pivot to the Radar. Yeah. Yeah.
What do we got there? So the Radar, and the key purpose of it is, is to take the community that is using these tools, the community that's very familiar with them, and give those who are maybe either less familiar or looking to engage with what is the best solution for my observability, IEC kind of issue. They can see what those who are familiar are using and say, well, adopt means the general community recommends this one.
Assess is they're still getting strong recommendations, but maybe not as high, but it may be your stack it works better or your use case it works better. Yeah. And then trial is just a bit less.
There's not as strong recommendation or maturity or reliability. Might be more niche. Yeah.
" Yeah. So yeah. Cool.
Very cool. Anything surprise you with the Radar? So I'll say, in the storage section, the Dragonfly, and there's one other project in there that I'm blanking on, but they're both Chinese-centric projects that I see very large adoption in China.
They're very mature, very stable. But in the Radar report, it came in in the, I believe it was the trial. Yeah.
Yeah. And that was just because we couldn't get a good sample size from China. Yeah.
Normally our sample in China varies from about 10%, 15%. This wave, because of Monte Carlo, we ended up with 1% of Chinese developers. Yeah.
Which meant we just missed them, and therefore what they liked just was missed over in the report. But so it's now something we're paying attention to, make sure we always get our 10%, 15% China because they're a really key part of cloud native. So we need to make sure they're involved.
It's a numbers game, guys. Yeah. Yeah.
It's that simple. Yeah. But it's interesting, right?
Because you look at, what is it, seven of the ten leading open-weight AI models are Chinese. Yeah. Yep.
But yet when you look at, okay, who's using these open-weight models? Silicon Valley- Yeah ... is probably the biggest user, right?
Yeah. So though they are perhaps made in China, as most everything is anyway, though they're made in China, because they're open-weight and they're being used all over the world, you may get a read from developers in other countries- Mm. Mm-hmm ...
who are using these models with these things. So it's not a substitute for going to the source. Yeah.
But it's something. Yeah. How many years have you been doing this?
So this is our fifth edition of the Radar. So we do it every six months, so two and a half years of the Radar. Yeah.
It did exist beforehand, but it was focused mostly internal to CNCF's community. Yeah. And this was more than five years ago, so it's definitely out of date.
Yeah. So we do it every six months. We have a pretty good rhythm of how it works, and we've been working with CNCF on cloud native data for, I think, three or four years at this point, supporting and understanding of that.
And CNCF has been a great partner because they know the space so much better than basically anyone else in there. So we have any questions, we can build it out. And I think Linux, the LF itself has been working with us for more than a decade.
So a long-term relationship we have between CNCF/Data. So yeah. Very cool.
It's really cool. Hey, we're almost out of time. People who want to go download the report- Mm-hmm ...
where can we go? io in, we have a little section under... Sorry, it's on the main page.
io. Yeah. co for additional research on the topic and area.
Yeah. Fair enough. Yeah.
Guys, if no one's told you before, let me tell you. Thank you. Thank you for the work you do.
Those reports are valuable. Thank you for having us. As a media journalist person, I rely on them.
Yeah. But there's a lot of users here as well. Thank you.
Thank you very much. Thank you. Hey, we're going to take a break.
I think I got one more interview in me left, and then we'll wrap up this version of KubeCon. Stay tuned. You're watching Techstrong TV.