Is Software Engineering Dying in 2026? (What the Data Actually Says) transcript

Junior developer job postings are down somewhere between 60 and 70% since 2022. And at the same time, the Bureau of Labor Statistics is still projecting for software developers over the next decade. That's about five times the average across all industries. So which is it? Is software engineering dying or is it growing? Now both the numbers are true at the same time. And if you understand why, you'll know exactly what to focus on and what to stop wasting your time on.

If software engineering isn't dying, but it is changing faster than at any point in my career, and the people getting hurt right now are mostly making the same mistake. They're optimizing for the version of this job that existed three years ago, not today. So let me break down what's actually happening. And then the four things that matter the most right now, if you want to become a software engineer, let's dive in. Okay.

So what's actually happening? Well, first I want to give you the owner's picture. And I don't want to sugarcoat it here. Around 77% of professional developers now say they use AI coding tools. And these tools are genuinely good at a specific category of work. Now that's boilerplate. Simple Crud endpoints, basic bug fixing test, scaffolding, routine, front end work. You get the idea. Now, that category is exactly what junior developers used to get hired to do.

It was the bottom rung of the ladder, and AI kind of just soared all of that off. Now, Stanford researchers found that employment for developers aged 22 to 25 dropped nearly 20% between 2022 and mid 2025. And once we have the data for 2026, I expect that's going to be even higher now over the same period, employment for workers over 30 in AI heavy roles actually grew, so the jobs didn't disappear. They simply moved, and they actually moved up the stack away from writing routine code and towards things that AI can't do.

So effectively. All of these junior jobs moved over to more AI related roles that are more for mid and senior level employees. Now, even the companies that are betting on junior developers are proving this point. IBM announced that they're tripling entry level hiring, and their reasoning is that the junior role is shifting away from routine coding towards judgment and customer facing work. So everyone on both sides of the debate here agrees on the same underlying fact.

The fact is that this job is being redefined. So let's talk about what it's being redefined into. Because once you understand that, you can know how to navigate this market. So the first thing that actually matters here is owning systems instead of just writing lines of code. AI has made writing code cheap. Now this is a big deal because in the past, most of what you did as a developer was actually write code. You had to mononymously sit there writing thousands of lines of code, and your typing speed genuinely mattered.

How productive you were in IDE was actually a selling point, and most junior developers spent most of their day writing boilerplate test scripts and actually just entering values into the computer. Now, at this point, that's no longer a skill that you're being hired for. However, you are still being hired for understanding a system. So knowing why the architecture is shaped the way it is. What breaks when you change something, How the pieces are talking together and what the data flow looks like.

Now that knowledge is what companies are really paying for, especially now. So double down on things like system design, taking a vague requirement and turning it into components, trade offs and making decisions that you can defend. You also want to double down on debugging real systems, because when production goes down, an but a human is the one who has to diagnose this and ultimately come up with the decision. And then of course on reading code.

Because honestly, right now, reading code is actually more important than writing it, especially when AI is generating so much that the bottleneck is actually how quickly you can review. So if you're c, you understand pull requests. You know how to quickly get through a code base, navigate different commits, go through various files. That is going to level you up now. Today, a senior engineer with these skills plus AI tools is shipping at multiples of what was possible just a few years ago.

And that's why their employment is growing while routine coding work is shrinking. So the leverage has really moved to the people who understand the systems. And I don't want to use the same word, but understand their leverage, how they can get the most amount done with the fewest resources. Now let's talk about the second thing that matters a lot now that is using AI tools like a professional. Now there's a right way of doing this and there's absolutely a wrong way of doing this.

Now the wrong version of this is vibe coding your entire job. So you accept whatever the model spits out and then you just ship it. Now that's how you end up with code bases that nobody understands. And that's what companies are already getting burned by. A bunch of junior developers coming in with vibe coding knowledge is not going to last very long. Now, the professional version of this is treating AI like a very fast junior developer.

It's something that you can supervise, that can get a lot done, but that you can't implicitly trust. So you use copilot, cursor cloud code, whatever fits your stack to move quickly. But you review everything, you test everything, and you never merge code that you could not explain. Now, this is actually something employers are screening for explicitly. Now they want people who are fast with AI, but also skeptical of what it's producing.

And to be clear here, being good with AI doesn't make you special in 2026, but being bad with it absolutely will filter you out. So it's no longer oh, I know how to use AI tools, so hire me. It's like you better know how to use AI tools. You're not even going to make it to the second interview. Now, the third thing I want to discuss, and honestly, one of the biggest opportunities on this list is learning to build with AI, not just alongside it.

Now, using cursor to write code faster is AI helping you build software? But what I'm talking about now is software, where AI is part of the features. So apps that call language models through APIs, systems that answer questions over companies, documents, agents that automate multi-step workflows. Basically, every company right now is trying to ship AI related features, and the number of engineers who can do this well is still really, really small because it's relatively new.

Now, that gap is why AI engineering roles in the United States are averaging somewhere between 100 and 70 5 to $200,000 per year, and why this has been one of the fastest growing skills in today's market. And this is not a career restart here, right? If you can already code, this AI layer is a set of learnable skills that sits on top of all of the skills you already have. So things like calling model APIs properly, prompt engineering, structured outputs, Rag vector databases, right?

Actually designing systems and having agent orchestration evaluations, observability so you can tell what's actually going on. None of these things require a PhD or a heavy math or genius level IQ. This is engineering. It's easy to learn relatively if you have the right path. And these are the skills that people are getting hired, literally hundreds of thousands of dollars per year. For now. The obvious question at this point is how do you actually learn that layer, especially if that's one of the biggest opportunities, because watching videos about it, even including this one, is not going to get you there.

No studies show that when you passively watch tutorials, you only retain about 20% of the material. But when you learn actively by actually writing code and building things like projects, your retention jumps to 75 to 90%. Now for a shift this important for your career. That difference is literally everything. Now that's why I partnered with Data Camp for this video. Now, I've used their platform for years to sharpen my own Python, an AI skills way before they ever sponsored any videos.

So this is something I can actually recommend from my own legitimate experience. Now there are so cute! AI engineer for developers track is built for exactly what we just talked about. Someone who can already code and wants to add the AI layer fast. Now it's 29 hours of hands on learning where you build real applications like chatbots, semantic search, recommendation systems using things like OpenAI API, Lang chain hugging face, pinecone vector databases.

You get the idea. Now it covers the production side as well. They have things like LM ops, which most people actually end up skipping. That goes over rate limiting API errors, structured model outputs so you can build reliable systems. And this track was actually just refreshed in July of 2026. So the content is current which really matters in this space. Now once you finish that there's an AI engineer for developers associate certification.

And it's not just multiple choice. There's a four hour practical exam where you build a real working AI application. So it certifies that you can do the job and not just talk about it. Now, the first chapter of every course is free. So you can start course one today and by link in the description will give you 25% off when you decide to upgrade. Now, if everything in this video is telling you the is moving, this is the most direct way that I know to actually move with it.

Okay, so let's move on to the fourth thing that matters, which is proof, right? In a market that's this competitive claims are worthless and evidence is literally everything. The strongest evidence is a shipped project. So either 1 or 2 projects that solve a real problem, end to end, that are deployed live with a link that someone can actually click. Now ideally this is with an AI component to connect that AI layer that we talked about, because that's what a lot of these hiring managers are scanning for and actively hiring for.

The most important part here is that you build something practical that real people actually use that's deployed. If you do that, even having one project is going to be ten half assed projects that don't solve a legitimate problem. Now, if you're not looking for a job and you're already employed, the highest leverage move that you can possibly make is to become the AI person on your team. So volunteer for the AI feature that nobody else wants to own.

Ship one real thing with a model that's actually inside of your current job, and have something on your resume so that you can prove you know what you're doing now. An internal move into AI work is almost always easier than landing it cold for mode side, and you get paid while you do it with pretty much zero risk and it gives you the evidence. Then go land something better at another company. I've said this many times before, but it's always easier to shift internally and then once you have the evidence, move to something better.

Okay, so those are what you should worry about, but what should you skip? So first thing, stop grinding LeetCode as your entire interview strategy. Now interview prep matters. You should do a bit of leetcode, but algorithms alone are the old game, right? Yes, companies are still testing them, but they're increasingly testing practical building and your judgment instead. So stop chasing every new framework, because the framework stuff is going to change constantly and really focus more on the fundamentals.

So system design, data, APIs, testing. Right. These things have never mattered more and they're not going away. Now also you want to stop doom scrolling. All of the coding is dead headlines. Now look, I'm a YouTuber. I'm guilty of doing it as well. But generally speaking, this is not something you can control unless you're going to completely change industries. You're kind of stuck going into software engineering, so all you can do is play the game to the best of your ability.

Focus on what's actually changing in the industry right now, and adapt just like everyone else has for the past 30 years to land a job. I know software engineers now that have never been paid more than they have before, and that's simply because they saw this coming. They adapted and they focused on it early. One thing I can tell you 100% is that nobody cares. Nobody feels sorry for you. If you want to complain and moan and just worry about the industry, you can do that, but you're going to get left behind with all of the people who don't do that and just take the steps to actually better themselves.

Software engineering is still a fantastic job. There are so many great benefits. It is one of the best jobs, in my opinion, in the world, even if we compare it to pretty much everything else. And while it has gotten more difficult to get a job, you still absolutely can. And the only way you're going to do that is by taking accountability in action. Again, it's kind of some tough love, but that's all I can say. And the people that have done that, pretty much everyone I know has been able to find something.

It just took maybe a little bit longer than it did in the past. So yes, software engineering is changing. It's been changing drastically and especially now. But you know what? It's changing into system design, ownership using AI tools, the engineering layer building prove all of these things have been changing for the past 3 or 4 years, and if you've looked into them, you've been able to get into a better role where you're probably making more money than you ever had before.

So I hope this video isn't dissuading you. All I'm trying to say is that there are certain things to focus on, certain things to not focus on. If you follow the guidance here, I hope that that's going to help get you back on track. Let me know what you think of the industry and all of the changes in the comments down below, and I will see you in the next one.