Kristina Kondrashevich, site reliability product manager at Electrolux, remembers the impact Kelsey Hightower made on her work with clarity.
“We attended KubeCon 2023, where Kelsey Hightower delivered a talk around open source projects,” Kondrashevich tells The New Stack. “I still have notes from that day, which we followed for building and open-sourcing our developer platform, InfraKitchen.”
“When we saw that Kelsey was going to be in Amsterdam again, we registered for the KubeCon 2026. We asked him for a photo, and I said he inspired my team and me to be brave and go [open-source their project].”
“He said to find him after his talk and do a demo for him,” she continues. “He challenged InfraKitchen with different questions, but was eventually impressed and approved what we do.”
This story isn’t a one-off for Hightower, who is known as the first big Kubernetes and cloud-native evangelist, as a charismatic speaker, and as the author of Kubernetes the Hard Way. But to hear developers like Kondrashevich tell it, he’s best known for the time he spends with others.
So what was on his mind at KubeAuto Day Europe 2026, a co-located event with KubeCon + CloudNativeCon? Not surprisingly, the impact of AI. On open source, on your codebase, and on your career. And perhaps, his rumored retirement from the tech stage. (Hightower retired from Google in 2023).
What does open source in the age of AI mean?
Following the deprecation of Ingress NGINX, so much of this year’s KubeCon discussion centered on tactics to encourage companies not to ignore projects they depend on. This includes maintaining a software bill of materials, especially for your open-source dependencies, and a continuous effort to support project maintainers and contributors. But this year it wasn’t just budgets to fight for. There’s a recent argument that, with the cost of creation becoming nominal with AI, you can just build your own instead of depending on open source.
“If they won’t contribute to open source and maintain open source, they have no chance with this [AI] stuff.”
“If they won’t contribute to open source and maintain open source, they have no chance with this [AI] stuff,” Hightower tells The New Stack. “You have a thing where a community has given you the biggest head start ever. Lots of other people are using [open source software] in production, and there’s an industry behind it.”
Anything you do generate with AI, he contends, will be half-baked at best. Then you’ll end up having to maintain it on your own, most likely leaving it neglected. Companies may try to take this AI-generated route, but will retreat, he predicts, at the first sign of a security flaw.
“This is why there are only so many recipes. It’s why people make scrambled eggs roughly the same way,” he continues. “Humans are a community. Community turns into culture, and most people want to be a part of something.”
KubeCon’s record 13,500 attendees signal that the AI era is drawing people closer to the open source community, not pushing them away. But growing dependence raises an old question: Is any of this sustainable? For two decades, open source advocates have struggled to convince leadership to fund what’s still often dismissed as “free software.”
“Open source underpins some of our commercial projects, so in those cases, it’s kind of a direct one-to-one to revenue — we make money off of this,” Hightower says, adding that he thinks companies should allocate a percentage to ensuring the open-source and Postgres communities are supported through hires and database maintenance.
“I do think every enterprise needs a little reminder; we are getting really far off the back of others, and if they were to deprecate it like the Ingress controller from NGINX, then look, lots of enterprises are scrambling now, and not one of them that I’ve met has thought about maybe they fork it and step up to maintain it,” he says.
“I think a lot of people have to remind themselves that open source wasn’t about getting software for free from someone else. It’s also about stepping up to maintain it when the time comes. And that’s much easier to do if you’re actively contributing along the way.”
This doesn’t have to be a very senior developer, either. Taking junior developers or even interns under your wing enables you to support an open-source dependency early and often, which is much cheaper than delaying OSS patches or migrating away if that project were to shut down, he says.
“It’s always been the same question. It’s always been the same answer. Nothing’s changed. AI didn’t change that question. You should have asked that question when you first started: How can I be better at this? And if you aren’t sure, you go find someone that you thought was better at this and learn from them.” – Kelsey Hightower
“Their objective is to return profit to shareholders, and if you have to burn down the rain forest to do it, that is the objective. So let’s not pretend there’s another objective,” Hightower says of AI growth at all costs.
How to be resilient in a tough economy
Tech workers have the same opportunities and challenges as ever. If you cannot speak to business, you will lose. The only thing that’s changed is the urgency of it all. So how can engineers, in the face of more AI-induced isolation, nurture those core business skills?
“If we’ve democratized your ‘hardcore’ skills, and now those are no longer as valuable as they used to be, then you’ve got to go learn some additional skills. So I think a lot of those people are going to get forced to broaden the scope of what their abilities are,” Hightower says.
As always, engineers have to be in a state of continuous curiosity and learning.
“It’s always been the same question. It’s always been the same answer. Nothing’s changed. AI didn’t change that question,” Hightower promises. “You should have asked that question when you first started: ‘How can I be better at this?’ And if you aren’t sure, you go find someone that you thought was better at this and learn from them.”
“Everyone is a junior engineer when it comes to AI”
This is even more important when you’ve been at the same place for five, 10, or even 20 years, but suddenly your job is at risk. You need to constantly be connected and ask colleagues in tech, even at different companies, what you need to know right now. Because, in a way, everyone is a junior engineer when it comes to AI.
“If you don’t know where you stand in the industry, then you’re not competitive. You are competing, you realize, with everyone else that is making progress,” he says.
“Take the hype, leverage your experience, and figure out, do you need it now? Maybe your company does need some of these AI tools. My guess is, go find someone actually using it in production,” Hightower recommends. “Can I get 30 minutes with you? I have some honest questions. What breaks? Why should I not use this? And I think if you keep doing those things, you’ll be much better off as an engineer.”
What if you see boasts of 10x productivity increases while the website or the product looks the same? He specifically warns against these many early braggers.
“Learn the patterns,” Hightower says, because while AI is good at patterns, so are good business-centric engineers. “Take notes. Be the person who says: ‘Hey, from experience, this is what works. This is what doesn’t work.’”
And, as he emphasized in his fireside chat at KubeAuto Day, keep questioning: “Is it worth it?”
“Who do you think trained the models we did? These are our ideas. They’re mimicking our creativity. So when I write the docs, and I write the code, those are my experiences being committed and serialized.”
Is there value in being full-stack anymore?
“I think some people hope that AI becomes this magic sauce you can rub on your YAML files and user experience pops out,” Hightower says.
But in an era of widespread AI, the software or DevOps engineer who can understand the entire system may be more valuable than ever: “It’s important that if you’re going to manage these systems, you need to know how they work. Does every developer need to know how to run Kubernetes the hard way? Absolutely not. But if you’re a practitioner and it is your responsibility to make sure Kubernetes actually works, that [is something] I think you need to understand. “
Specifically, he argues that while everyone says they care about security, many engineers are standoffish about addressing it, which is in part why they outsource it to external tools. But, especially with a surge in AI-related vulnerabilities, the security tools are becoming vulnerable.
“A lot of people have said they have lost the ability to understand if something is secure or not. What use are you at that point?” Hightower says. “This is the danger of any system that removes people’s understanding, because when it’s time to understand, you won’t be able to.”
Whether it’s contrary to or in support of the claim that engineers will be doing less engineering in the age of AI, Hightower still firmly believes folks should be thinking about building the whole stack the hard way at least once or twice.
In the end, while Hightower seems more suspicious than most of the ROI in AI, he does still seem to apply his characteristically optimistic lens:
“2026 isn’t the deadline for all human endeavors, all human experience,” he says. “There are species we still haven’t discovered because we haven’t yet gone to the very depths of the ocean or to the edge of the universe. It’s not over yet. This is just a checkpoint.
“And don’t think that means we stop thinking. I hope you won’t stop thinking, even if everyone else decides to stop.”
YOUTUBE.COM/THENEWSTACK
Tech moves fast, don’t miss an episode. Subscribe to our YouTube
channel to stream all our podcasts, interviews, demos, and more.
