AI Fuels Demand for Open Source Tech Talent

Hey everybody, we're back at the Open Source Summit in Minneapolis. I'm here with Clyde Siepersad, who's running research these days for the Linux Foundation, among other things. But you have a new report out.

We do have a new report out. We've done this for 20 years. It used to be called the Linux Jobs Report and then the Open Source Jobs Report, and then we realized all tech is open source, so now it's called the Tech Talent Report.

Okay. And we're breaking through some of the received wisdom, especially around AI coming to eat all our jobs. So conventional wisdom has it that that's exactly what is happening, and we see all these layoffs that are ascribed to AI, but the research shows something almost the opposite is happening, and that there's more demand for people who have tech experience because of AI.

Well, I think it's the sort of truism that reality is complex, and if anybody reads the headlines, it's obviously true that there's been over a half a million reductions in positions. The second part of that sentence is at the largest tech employers. So when we did the research and we looked at the data, sure enough, the largest employers, when we drew a cutoff line at, I think 20,000, above 20,000 employees, they had in fact reduced their staffing in the technical staff.

It turns out that the small and medium enterprises, which are like 75% of the employment base, increased their staffing, in large part by hiring the half a million people that were laid off by the FAANG group. Yeah. And so nobody runs out and writes a headline that midsize company hires 10 developers, but they write a headline saying XYZ laid off 15,000.

Mm-hmm. And so yes, the news cycle talks about big layoffs happening at the very largest tech employers. The reality cycle is that most of these people are getting placed typically within two to three months at a smaller organization, who, by the way, was probably trying to hire them during the 2021 to 2025 period where the big tech companies were sucking up all the talent.

Yeah. And so two things can be true at the same time, which is high levels of highly visible reductions, but also an underlying trend of lots of companies investing because they don't want to get left behind. And if I look at the big tech companies, it seems to me they have now reduced their hiring to what it was pre-COVID levels, and they basically, to your point, went out and hired everybody and anybody they could find on some assumptions about the way the world was going to permanently be.

Right. But it looks like they're now just kind of resizing to what they used to be, and that frees up a bunch of talent for everybody else. Is that a fair assessment?

I think that's exactly what we're seeing, right, is that the talent is getting to go to other places. And arguably to places where they're smaller, they're more nimble, and they can actually take advantage of some of these advances in terms of using the foundational models and implementing agentic workflows. I think you're going to see a lot of creativity coming out of that, because it's generally been true that smaller organizations tend to be nimbler.

They tend to have less business that they're trying to defend, and they're trying to be more willing to experiment with sort of radically different ways of doing things. Do you think that they will hire people where they find them, or are they going to ask people to move to where they are? Because there has been this whole conversation about remote work that's been going on for the last four or five years, and a lot of the talent that's now available was in places like the Bay Area or New York and maybe not in Michigan or Kansas, where a lot of these middle-level companies are.

Well, yeah, I think one of the things that you're seeing, and this wasn't in the data that we collected here, but I think you see anecdotally, is those smaller employers tend to be in the community where the person had moved to when they left the Bay Area to get away from $10,000 a month rent. And so there's been this sort of interesting balance of smaller, midsize companies hiring in the not San Francisco, not New York, not Austin, not Boston type places, where folks moved to kind of during COVID and when it was possible to do remote, where they find that they can afford a house and actually put their kids in good schools. So I think some of it is in-person, but it's in-person in communities where people previously weren't able to even hire folks in these roles.

And then also some of these smaller companies have gotten more used to being able to work with remote employees because they had to do some of that kind of during the pandemic. So I don't think it's one narrative or the other. I think you're seeing a blend of people moving away from the super expensive tech cities and are now working with places that are actually close to where they work, but also small and midsize companies have had to get more comfortable with the idea of having certain people not necessarily be on site all the time.

In the report, are there particular skill sets that folks are looking for more so than others? Is there anything in there that kind of would be an indicator to somebody to say, "Concentrate on this particular skill"? Yeah.

When you look at where people feel understaffed, the biggest jump year over year was cybersecurity. We actually had the survey in the field before Mythos came out. I think if we had been a month later, it would have been an even bigger number than the 57% or whatever it was.

But then right behind that was some of the greatest hits, right? Cloud native stuff, DevOps stuff, agentic workflows. And so I think there is this big cross-section of skills becauseWhat companies are increasingly realizing is the shiny object is the agentic workflow, and that's what they all want to grab onto.

But the reality is, to get that, you actually need a whole set of infrastructure that's lined up and available to be exposed to an MCP server, and with the right configuration of permissions to allow an agent to come in and talk to it. And so in order to get that stack, that sort of agentic-ready stack, there's a ton of work you need to do on your legacy systems to- Oh, no ... upgrade it, improve all these different components, right?

And so it's not pick a level and do that. It's pick a level and then discover that everything else needs to be upgraded. So you have to upskill people to do cloud native and microservices.

You have to upskill people to do proper DevOps review time with a high agentic input in the sort of review cycle, and then you have to do the stuff on the agentic workflows and the MCP servers and all the other bits. Are you seeing those mid-level organizations willing to invest in the training to drive that? Or are they kind of expecting that the people that they go hire will have those skills, or they're putting most of the weight for getting those skills on the individual employee?

Well, what's been interesting is across the board, organizations year over year, so in '25 it was higher than '24. In the new report for '26, it's higher than '25. Companies have sort of found religion on the importance of upskilling.

Mm-hmm. So it used to be, as recently as four or five years ago, that when you talk to hiring managers about their needs in new technical areas, their overwhelming preference was hire. They wanted to go outside and pluck somebody off of LinkedIn that had Kubernetes skills and that had done this somewhere else.

And now I think a lot of that has-- People realize that it's incredibly hard, it's incredibly expensive, and it's eight months from the time you open the req to the time the person has been hired, onboarded, and starts to be productive, and 40% of people leave within the first year. And so, you're grasping at air- Mm-hmm ... if that's all you want to do.

So I think people did spend time coming up to the reality of what it means to try to bring in talent. But I think they also realize now, as they've gone through the initial experiments with agentic, that the value of the implicit knowledge of the people who built and ran the current systems is irreplaceable, that you just can't parachute somebody in from the outside. And so, the people who remembered that, we tried this five years ago, and it went horribly sideways because of XYZ reason, or these customers revolted because some aspect of how we interacted with them, that stuff is incredibly valuable when you're trying to put agentic workflows in place because those are the things that are not written down.

They're not in the data. They're not in the training corpus, and the model is going to be completely blind to them, and it's going to make all the same mistakes that everybody collectively has wisdom on. And so there's this increasing, I think, realization that the people who run your current stack are the ones with the knowledge to help you build an efficient, effective stack on the new technologies because they're the ones who know where the bodies are buried.

They're the ones who have all that implicit knowledge, and it's actually not that expensive to upskill your existing people, and it's not that time-consuming to do it, versus $40,000 for a headhunter to find you the most whiz-bang agentic programmer who probably is going to get a better offer six months down the line and move on to the next thing. The second thing that people are talking a lot about is that there's a hollowing out of the entry-level job positions, and what seems to be the concern is that AI is automating a lot of the tasks that we used to hire kids for from their entry-level job, and then we won't have enough experts because eventually the current experts will retire, and there'll be this gap. Did you see anything like that in the research, or what did you actually find?

So, even outside of the research, it's prima facie true that it's harder to get a job coming out of college with a computer science degree than it was over the past several years. Part of that is we have this cultural moment where employers were rushing onto campuses and doing job fairs, and kids were getting four and five offers, and it was sort of like breakfast buffet. It's just like pick and choose, and the employers were coming to us, and that music has stopped.

And so we have a generation of folks who their expectation was that jobs were going to come to them, and the reality is they have to go to the jobs, and they have to figure out how to embed themselves in their local communities, which startup pitch competitions to go to to see who the hot new companies are in their area, which friend of their uncle that works at a company can get them an introduction. And there's a lot of resistance to that because it feels like, "Oh, I'm not a nepo-baby. " And the reality is, people are being more circumspect about hiring net new talent.

It's been interesting to watch. I'm sure you see this. People are almost sheepish about hiring new employees.

It's sort of like, I don't want to be the weirdo that still has to hire new people because everybody else is in this agentic wonderland where all the work just happens auto-magically. And so there's this weird cultural bias now against hiring new people, right? It's sort of like you're the weirdo if you want to bring new talent in.

But just like with every other technological wave of innovation, who are the people most passionate and most invested and most exploring and least attached to yesterday's ways of doing things? It's always the youngest people. " You would want to bring those people in, but you'd want to give them a different work experience.

Instead of locking them in the basement and making them write basic code for 12 months before you gave them real responsibility, you'd want to give them a cross-section of responsibilities. And that's partially on the companies to design the jobs so that they're taking advantage. " What dev stuff do you have to do?

They have to go experience, learn what it is like to run pipelines. They have to go learn, experience what it is like to run an inference layer or to feed data into a model or to think about security. And so I think we are asking younger folks to expand their breadth and get more exposure to these different concepts and not be like they are in the headlines.

And we're asking them to go do more investment in finding the right opportunities and building the relationships, because the number one thing I hear from folks who hire young people is it was somebody they knew. It was somebody who interned for them. It was somebody that got introduced to them that they got to know, which is actually not surprising because anybody who's posted a job in the last year wakes up the next morning to 10,000 AI slop applications, and their first instinct is delete all and start asking my friends.

That's the next- Who's a real person that might actually know something? Most of the senior level IT folks that I meet, at some point, were hanging around an open source project. They were contributing to that.

That's how they built their network. That's how they got to know people. Is that something that the next younger generation has to relearn and rediscover?

I would hope so. All the cool stuff's happening in these open source repos. Go where the fun is, go where the heat and the excitement is.

" Back yourself. Do something more ambitious with that code release because you actually have access to tools to do a bunch more sophisticated coding than maybe you'd be comfortable doing as a newbie. But you've got to back yourself that you could have impact, and you actually have the tools to help you do that in a more sophisticated way.

So what does the future of this market look like to you from a jobs perspective? And if we were going to be looking at this study a year from now, what do you think it'll show you? I think it will show that all of us have to be upskilling all the time.

And by all of us, I mean all of us, not just young people. Mm-hmm. And the thing that became true, I think, with this foundational model and coding agents, the all of us now includes more people who weren't technologists and who were business side folks and who are having to get into it and start to think about, can I do some coding?

Can I get better at understanding what these agentic workflows are, what the implementation looks like? Because I think it's jumped the shark. This line, it used to be that the devs were here and the engineers were here, and maybe there were some QA people in the middle, and we've already collapsed all.

That went away with the DevOps thing. Mm-hmm. But there was still this line between the business people who are over here and the tech people who are over here, and I feel like that's what's breaking down now.

And so the tech people for sure know that you have to keep building your skills. Yeah. There's no resting on your laurels.

But I think now even the business people are realizing that you can't just not pay attention and not even attempt to get familiar and begin to experiment with this stuff. I think that's what's different. The new normal is all of us are learning all the time, and nothing matters that you learned five years ago because it's just, you know.

I feel like one of the things that we don't talk enough about too is that all these organizations, their ability to consume tech was limited by the talent they had who could manage it. " Right. In the age of AI, we should be able to manage more with a limited or same pool.

And so does that mean at the end of the day that we'll see just more consumption of IT in general because it'll be easier to do at a higher level of scale and organizations will simply try more things? We don't have to go very far back in the way back machine to remember when people were saying, "One engineer can manage a million servers. " This was the cloud and DevOps revolution.

What happened? We ended up with more engineers because the appetite grew. There was a great example this morning in one of the keynotes where the guy on stage had done a pull request against a feature request from 2019 because it just sat there since 2019 and nobody wanted to work on it.

Same question of too much work, not enough people. Mm-hmm. Well, now we're being able to get more done with individuals coding across a bigger footprint, and maybe we'll start getting to those use cases which are still valid seven years later, which nobody ever worked on.

Our appetite for using technology continues to accelerate. " The paradox has kicked in, and our usage grew so much that we actually ended up needing more people running it, imagining it. Now, maybe that's not true if we get to this true AGI, artificial general.

But for right now, there's no cap on the appetite for doing more stuff. Right? And I think we're just going to see more and more stuff get moved into workloads.

And in particular, with some of these smaller companies, they want to do stuff they just haven't been able to. It was too expensive. They couldn't get to the staff.

The technology wasn't mature enough. It felt like it was risky. So I think you just get this incredible unleashing.

And I had a doctor's appointment last week, and I was watching the person key the stuff into the charts, and I'm just thinking, "Wasn't this transcribing the discussion we were just having? " Mm-hmm. There's opportunities everywhere, Mike.

Everywhere you look, there's opportunities to capture, digitize, move stuff into workflows. Every time we see a transition or a shift of this magnitude, the conversation inevitably comes back to, well, do I need certificates or what? Do I need training?

And if I have the training or if I have the hands-on, do I need a piece of paper that says I should? Right. What's your take on what should folks be doing about their careers and the certificates and what that means?

Yeah. I think there's plenty of use cases where people have a body of work that they could point to, and the certification piece is not as important. I think what's different right now is the speed at which the technologies are entering the workforce is so much higher.

It's really hard to say. Nobody has two years experience implementing agentic work queues, right? And so some of the stuff we used to rely on of, well, let me show you my work.

Let me show you my GitHub profile. There's just not enough time, so you have to have other ways of demonstrating that you're comfortable, that you have facility with doing it. And so I do think there's a lane for certifications because, A, they're not I need to go to school for a year and come back and show you.

That's clearly not happening. Mm-hmm. It's also clearly not happening that wait for a year while I build up my portfolio to show you this nice code resume.

" Because what's the number one thing employers want is they want to trust that the person has the skill. Mm-hmm. And the old model was, well, you went off and did the training and the degree and you came back, so somebody vetted it.

Or I poached you from my neighbor down the street, and you did it for two years there, so now I have confidence. So where does the confidence come from when the thing is less than nine months old? Right?

There's going to have to be some way of signaling. I'm not saying cert is the only way to do that, but it sure is a convenient way to do it when things are moving super quickly. So as you look at all the data in the report, what's your best advice to folks?

What's that thing that leaps out and says, "Folks need to pay more attention to this"? I think it's two pieces, right? I think for the new folks coming into the workforce, it's get out of your vertical and learn as much stuff about the other pieces of the stack, because nobody can be just a dev, just an engineer anymore.

Those days are over. And so invest aggressively in yourself to make sure you have a nice wide aperture on your lens. And then the second is for the people who are already employed, and they're midway into their career, and it's a version of the same advice, which is you have to keep learning and pushing, and there's no such thing as I've made it.

Here's my 15 years. I'm onto the next thing. I think we're all in this model now where we all have to be lifelong learners.

All right. Folks, you heard it here. If you want to be in tech, there's one thing you always need to remember.

Always be learning. Always be learning. Thank you, Mike.

Thanks for being here. And we'll be back in a minute.