CNCF, SlashData on where cloud native AI use lies into 2027

This year's Cloud Native Computing Foundation (CNCF) State of Cloud Native Report for Q3 detailed how cloud native developers use AI from drafting code to autonomous backend execution.

In the report released on November 10, cloud native developers are more likely than their non-cloud native peers to use AI coding tools (75 per cent vs 55 per cent) and to integrate AI into their services (67 per cent vs 38 per cent). Among developers integrating AI into their services, 49 per cent report agentic systems in production. Cost, data privacy and security are the leading concerns among those who have not. 

In North America, the report found that 21 per cent of cloud native developers use AI to draft code before reviewing and editing it, while 12 per cent use autonomous agents for coding tasks. On a global level, 16 per cent said they set the goal, while an AI agent plans and completes development steps.

Bob Killen, Director of Technical Program Management at CNCF, said most cloud native developers adopt AI with a human reviewing the output rather than going fully agentic, and that infrastructure work in particular needs double-checking. 

Despite woes of AI replacing jobs in the tech sector, since last year's Q3 State of Cloud Native Report, the global cloud native-identifying developer population has grown by more than 5.5 million, from 15.6 million to 21.2 million people.

Despite the dramatic growth, SlashData, the Manchester-based analysis firm commissioned for the report, expects cloud native developer numbers to plateau over the coming months as adoption expands in areas such as back-end development; other developers in digital games, desktop, and mobile may adopt at a different pace.

Liam Bollmann-Dodd, Principal Market Research Consultant at SlashData. stressed that a plateau in the measured figure wouldn't necessarily mean cloud-native use had stopped growing; more people may rely on the technology without identifying themselves as cloud-native developers. 

"We basically filled out the whole of back end. We filled out a huge on the back end group that we're going to adopt cloud native, and then we start seeing the rest of developers that start picking it up at this level, and I imagine there'll be another level - there'll be something that happened that'll push another group of people to consume it," said Bollmann-Dodd.

Amongst this growing number of cloud native-identifying developers, Killen said human review remains the norm for infrastructure work. He said historical knowledge and tech debt still require people: a prototype can be built quickly but still needs review by multiple teams and consensus. He said a popular topic at the CNCF Maintainer Summit in Salt Lake City on November 8 was how to grow new maintainers in the AI era. The general consensus: humans are still needed to understand complex systems and to describe precisely what they want built.

"I don't think it'll decrease the jobs. People might try and do it. We've seen a couple of companies already that have tried to go full AI or do large cuts to then push more AI usage, but it winds up frequently tending to bite them because there are so many other aspects to it," said Killen.

Bollmann-Dodd agreed that the number of developers and jobs will increase overall. That being said, he expects the next two to three years to be less clear, as teams cut developers, realise the mistake, and rehire slowly after about six months.

"[Companies] are going to very quickly realise there's a reason why we've always had junior developers as a huge sway for our market," added Bollmann-Dodd. "There's a name for it in the Japanese economy, [Employment Ice Age], when there is a recession and a whole generation of workers who just never got jobs. It's going to be a generation of developers who just never really got on the on the footholds. And sure they can create magical things with AI, but you need a lot of advice in the industry. You need your first step to prove that you can be a developer, not just a Script Kid."

Bollmann-Dodd said cloud native developers skew more experienced and more sceptical. He said senior developers tend to want to check an agent's work, because small mistakes can break systems built around older approaches. He also said developers have narrowed the tasks they hand to AI but rely on it more for those tasks, after realising their job covers only part of the software lifecycle.

"In seeing some of the jobs that have come back on the market, like 'we're looking for a senior Cobol developer', and it's the amount of people that can still do that is a handful," Killen playfully added.

For now, as AI usage increases, so do those identifying as cloud native developers.

"We suspect there are more cloud native developers than we measure because there are people who don't know what they're using underneath of it," said Bollmann-Dodd. "Then the question becomes, what makes a cloud native developer? Is it just the use of technologies, or is it an engagement and understanding of what you're using? I think it depends on who you care who cares about it. I think for CNCF, knowing the number of people who are actively aware of what they're doing and building on these things is more important. But I think we're getting to the point where I'm looking into my next research wave of the number of people who are dependent on cloud native without knowing it. I think that number be even more. "