Fresh Data Has Us Asking, Does AI Demand Kubernetes?  🔗 **Source:** [Fresh Data Has Us Asking, Does AI Demand Kubernetes?](https://share.snipd.com/episode/0dac3c09-b08c-4659-a38b-4ff7e7acb3eb) ## Highlights - **Cloud Native Means Consumers Of The Stack** - Cloud-native now includes many developers who consume platforms built on Kubernetes rather than directly using it. - SlashData flags 19.9 million cloud-native developers by identifying use of containers, orchestration, and related backend tech rather than self-identification. Transcript: Bob Killen The interesting thing. The definition of cloud-native and a cloud-native developer has changed significantly. It used to be someone that was using Kubernetes, was using containers, but now much more of the space is people consuming the technology that sits on top of Kubernetes. So they'll be using it indirectly. Liam Bollmann-Dodd Yeah, so when we do the analysis, we don't ask people, are you cloud native? Because basically no one knows what it means. And we're pretty fair that we have to be clear about how we understand it. So we basically ask a bunch of questions about technologies used in backend or cloud spaces. And from those, we look at things like Kubernetes, container orchestration, containers, Microsoft, it's like a bunch of technologies people generally see as cloud native. And if you use enough of them, we say this person is probably cloud native. They may not call themselves it, but we would think they reflect someone who belongs in the cloud native community. And that's where we get the very large 19.9 million number for the latest reports. Jennifer Riggins It is a very large, but also probably 40% of what AI told me was the amount of developers in the world. So are the rest not cloud native? What does that mean? And yeah. Liam Bollmann-Dodd So from my perspective, I think we are like possibly undercounting the number, but then, but the way we'd be undercounting it is what Bob said, which is that there's a whole bunch of technology Built on cloud native principles of cloud native technologies that people just may not be aware they're using. They just not realize that this thing is all containers or all Kubernetes. And they're just not aware of that because they don't care. It's not part of their Rob description. So we're possibly underreporting a chunk. But we also like we are being like quite loose in the fact that someone who doesn't identify as cloud native, we will still give them that label even if we think because we think they're Using enough technologies. Good example, like GitHub Actions. They're all containers under the hood. Jennifer Riggins Yeah. Okay. So even if you were, but if you're using a SaaS product that also uses it, would that count or no? Liam Bollmann-Dodd So in our survey specifically, they may not be identified as that because they have to, we ask them like, which technologies are you using? And like, we have like done follow-ups and you can see that people like, they'll say they use Kubernetes. Then we ask them what they do. They have no idea what they're doing. They just know that, they just know they're using Kubernetes. But, like, it's part of, like, the chain. ([Time 0:02:46](https://share.snipd.com/snip/89794020-bbde-41da-b416-a41654d9c038)) - **Security Is A Tug Of War With Better Scanners** - Security tooling and scanning are improving alongside automated attacks, creating a dynamic tension rather than one-way deterioration. - Bob notes tools like scanners and controllers can suggest configuration hardening for Kubernetes clusters. Transcript: Bob Killen But we're also developing a lot more better tools at securing the systems. And a lot more of tools of being able to, like there's MCP server for Kubernetes. So they can go in there and look at it and like, it'll make a bunch of suggestions to lock down the config. ([Time 0:06:28](https://share.snipd.com/snip/325e735f-7745-4603-bf4a-f11a33d70f87)) - **IDPs Contain Risk From AI And Junior Developers** - Internal developer platforms (IDPs) let organizations lock down infrastructure so AI or less-experienced developers can't break production. - Platform teams build pipelines and security controls centrally, enabling developers to focus on product work safely. Transcript: Liam Bollmann-Dodd And I also think one of the things mentioned about security and like the kind of safety with AI making things like better and worse at the same time is that one of the approaches you can Take is if you take the kind of developer platform, internal tooling, where you can basically prevent people from being dangerous to themselves. Like you can control everything at your end. All the security is done by someone who actually understands how security works. All the pipelines are built by people who actually know how pipelines work. ([Time 0:09:10](https://share.snipd.com/snip/6f356e9e-ef12-49c5-adbb-83c1a715aa86)) - **Hybrid Cloud Is Regional And Industry Driven** - Hybrid cloud adoption is driven by industry and regional factors like regulation and existing data center investments. - Financial services and regulated industries prefer hybrid to manage sovereignty concerns such as the US Cloud Act. Transcript: Liam Bollmann-Dodd It's regional. I think it's regional dependent. And I also think it's very org dependent. So like in our analysis, you do find that like fintech, insurance, hardware, they are really on hybrid. Like they're much, much higher rates than like a SaaS company or like an e-commerce company. So I think there is a lot of industry specific regulation that wants, like it just makes your life easier if you're compliant in your thing. And then you also have a lot of these really large companies where they don't want some services to be on the same like the US Cloud Act has done a number to how a lot of European companies Behave like they just really cannot afford to be on a on like the big hyperscalers because the US Cloud Act under the current understanding in the EU like is a risk to how the EU legislates Itself. But at the same time, you have what I think is missed a lot of time is to be in a cloud space, and we talk to cloud people, is that a lot of people just aren't on the cloud. ([Time 0:12:10](https://share.snipd.com/snip/82e106da-a1ec-43a2-b417-2e720cc5084c)) - **AI Can't Replace Maintenance And Compliance Work** - Relying on AI to replace junior developers or to rebuild open source is short-sighted because product maintenance and compliance are the real long-term costs. - Liam says generated prototypes often miss edge cases and the upkeep outweighs initial speed gains. Transcript: Liam Bollmann-Dodd I will say that in our data for the first time ever we did see a decline in the number of younger developers and less experienced developers and obviously the joke internally was are they Just too busy looking for jobs to do our survey was like it was obviously a comment we made and we don't know like we've you know the first with the first wave of the data we're not going to Begin like saying the sky is falling on one wave of data but it is worrying to see this and also i think what's more worrying is the the open source is building yourself with ai is a solution I don't think it's a real solution because like if you ever work at a sas company you'll realize that building a product building a service is like one percent of what it takes to actually Run a product or server. It is maintaining it. You could be working, make sure regulations comply. Like I could go home right now and go on Claude Code and it could spit out an equivalent version of Atlassian's products and it would break probably even a day because I would, there'd Be something missed or something different. Like that's just like one small example. So I think open source is being hammered because people are being short sighted at the moment. I think the industry at the moment is having a real cycle of short sightedness when it comes to so many layers of other approaches. I think I could give an hour long rant about why this is the case. ([Time 0:14:10](https://share.snipd.com/snip/fa38052c-b602-48dc-971a-0d2ebd2e8e9b)) - **Inventory Dependencies Before Replacing With AI** - Do inventory your dependencies and calculate the cost to fork or replace an open source or commercial component before switching to AI-generated code. - Bob advises weighing maintenance cost and compliance risk rather than short-term AI convenience. Transcript: Bob Killen There was a talk where a person was talking like, you know, they basically used AI to code something. And it wound up costing them a fair amount of actually like token usage and all that. When literally they just did a Google search, they would have found an open source project that was mature that would have solved their problems. So instead of actually like doing even a little bit of due diligence, they went to, you know, just generating a solution that wound up being worse than what is actually out there. There was another instance where literally they had set up an AI cron job to do something that literally could have been done by a one-line bash script. And so that was much more both computational and cost expensive for something that could have been done and much more maintainable by literally a one-line script. Jennifer Riggins It's very interesting also because companies that are looking to get rid of junior developers, they don't think they need it because of AI. Maybe they should be looking at their open, maybe they should have an open source software bill of materials. And then maybe they should be looking at where their risks are because that company could, that maintainer could. Bob Killen That is something that I have stressed repeatedly. They shouldn't pay juniors to maintain. Not enough organizations are actually doing the due diligence to see, to look at their risk. And even these days, looking at the risk of using some of the commercial software that's coming about. When I walk organizations developing a strategy, like the first thing we go through is what are you using? What would it cost for you to either fork and maintain, find a competing open source project or find a competing like commercial product and then trying to weigh the numbers of how much That would cost to switch to any of them. And that gives you an idea of whether like, okay, should we invest in open source project or should we pursue one of these other solutions? ([Time 0:15:23](https://share.snipd.com/snip/d09e9d14-d3ce-446a-abf9-44e7e684a78c)) - **AI Replaced A One Line Script And Cost More** - A speaker shared cases where teams used AI to code solutions that were costlier and less maintainable than existing open source options. - One example: an AI cron job replaced a one-line bash script, increasing compute and token costs unnecessarily. Transcript: Bob Killen There was a talk where a person was talking like, you know, they basically used AI to code something. And it wound up costing them a fair amount of actually like token usage and all that. When literally they just did a Google search, they would have found an open source project that was mature that would have solved their problems. So instead of actually like doing even a little bit of due diligence, they went to, you know, just generating a solution that wound up being worse than what is actually out there. There was another instance where literally they had set up an AI cron job to do something that literally could have been done by a one-line bash script. ([Time 0:15:26](https://share.snipd.com/snip/fc114dc2-cc56-431b-b51b-598da7e78cc3)) - **Detect IDPs By Tooling And Teams Not Labels** - Ask concrete questions about tooling and standardized processes to infer whether an organization has an IDP. - SlashData uses questions about standardized DevOps tools and internal platform teams rather than just the label IDP. Transcript: Liam Bollmann-Dodd So we do it in two different ways. We do it differently in the tech radar as we do in our big survey. So in the tech radar, we just ask them straight up like IDP. And the reason we kind of do that is... You say IDP or do you say internal developer? We say internal developer platform. Jennifer Riggins Okay, because there's also internal developer portal. Liam Bollmann-Dodd Yeah, so we say internal developer platform, and then we shorten IDP. And the reason we do that is we don't have enough questions to really get really granular about it. And these are people who say they use Cloud Native, and they say they've used these tools. So we kind of hope that they have a bit of a context for it. In our survey, our bigger survey, we don't ask them, do you have a platform? We ask them, do you have any of the following things in your organization? Do you have standardized DevOps tools? Do you have an internal like we call it an internal like platform like we use different words rather than just specifically saying IDP because we're all we're surveying like more than 12,000 developers from around the world with different languages with different like exposures. With those questions you can assert that they have an actual IDP or not. ([Time 0:18:14](https://share.snipd.com/snip/3ad25b2b-1d93-4e59-a2ea-daeaba4c7c46)) - **Open Source Contributions Landed A Google Job** - Bob described being recruited by Google because of his open source contributions to Kubernetes after two decades in academia. - Contributing publicly functions as a visible resume that directly led to career opportunities. Transcript: Bob Killen I can say at least when it comes to the job market and getting a job. So I was recruited by Google specifically for my open source contributions. What project? Kubernetes. Oh, okay. I was working in academia. I've been in academia for like 20 years and was contributing to kubernetes on the side as a passion project and literally got recruited that way and i know multiple other people that Have had the same path um a co-worker of mine george castro he has he calls it his his kids are a whole bunch of like young students and things like that that have been working on his product Bluefin, which is a like Linux desktop type thing. And they've all become like they were sending patches to the Linux kernel and they've like a bunch of them have now moved on to get like jobs in tech and good jobs in tech. So it's there and that's how for me how open source can open the door is it is a public resume of your technical acumen. ([Time 0:20:36](https://share.snipd.com/snip/22fc15b1-9e29-4f3a-86e7-7f1c51022e36))