If you're a Zoomer, this one's for you: Everything Gen Z needs to know about the 2025 tech landscape

If you are a person who exists on the internet—which, if you’re reading this, you are—you’ve probably experienced the same influx of “wrapped” marketing and commemorative digital artifacts that the rest of us have. While the campaigns have reached every generation of internet users, I personally feel it’s as Gen Z a thing as anything could be—I mean, it was originally ideated by a Gen Z intern at Spotify. And what’s not to love about it? It wraps everything you’d want to know about your year in one neat, aesthetic, easy-to-access place.

I spend a decent part of my workday at Stack Overflow writing for and thinking about the Gen Z perspective. So as Spotify Wrapped day came and I started to flip through the endless Instagram stories of people’s questionable but valid music taste, I got to thinking—what if I made a Gen Z wrapped for Stack Overflow? So here it is: everything a Zoomer needs to know about the 2025 tech landscape—from the AI bubble, to agents, vibe coding, and the job market—wrapped perfectly just for you. And if you’re like, “Congrats or sorry, I’m not reading all that,” don’t worry, I’ve added a TLDR; and links at the end of every section. Let’s get into it.

This was the question of 2025: is AI a bubble? And if it is, when will it pop?

If, like me, you are not old enough to remember the dot-com boom, let me give you just a quick rundown of what they mean by a “bubble bursting.” In the late 90s/early aughts (hey, that’s when I was born!), people started getting really excited about the internet. Brand-new companies started springing up everywhere, attempting to profit from the gold rush, and investment firms were dumping absolute bucketloads of money into them.

They called this the dot-com bubble—a pretty cool name if you ask me. At its peak in 2000, the hype around the internet resulted in $300 billion dollars being poured into these companies. And, oh boy, let me tell you, these companies were overvalued—Pets.com raised $82.5M in an IPO and went bankrupt nine months later. When the hype bubble burst and their seed money dried out, many of them went under, leading to a 77% drop in the Nasdaq index just two short years later.

This is the exact thing many folks worry about with the new influx of money around AI. Just this year $1.5 trillion dollars was invested into AI worldwide. I mean, that’s a lot of pennies. Plus, people are concerned about the circular nature of some of these investments, since all of the big guys are sort of just passing the same $100 billion back and forth through deals and investments. And the stocks are growing, but at what cost? Some are saying the U.S. is making one big bet on AI that we better hope we win.

So, is AI a bubble or what? Well, your intrepid Gen Z writer may not have the answer about if and when the bubble will pop—if I did, I’d be way more famous—but it is something to watch out for if you’re a young person early in your career. If and when the bubble pops, the following recession will be everyone’s problem. However, one of the important things to note for early-career professionals is that many AI startups are becoming merger and acquisition targets much quicker than they would have been in the past. According to Crunchbase, there’s been a 13% increase in global startup acquisitions this year. And they’re getting their pennies: the dollar volume for these deals has increased 115% in 2025.

For you, my dear Zoomer reader, this means working at a startup may lead to a shakeup pretty quickly in your career when the acquisitions come in—faster than they did in the past, if the above numbers are to be believed. While this isn’t necessarily a bad thing (it can be quite exciting, actually, and my GenX colleague says cashing in equity is nice), with the general uncertainty around the AI bubble and the job market for young people, it’s just something to think about as you’re looking at your next career move.

Ultimately, AI is the type of technology that’s readyset to disrupt everything, but most of all itself. We saw this pretty clearly when GenAI quickly moved into agentic AI in what feels like the span of a few months. What that means for us young people is yet to be seen, but I’m keeping my eyes peeled for that bubble pop, and you should too. If you’re looking for somewhere to start, we did a pod on whether AI is a bubble or a revolution with HumanX and Crunchbase earlier this year. Shoutout to our GOAT host, Ryan Donovan, for this one.

TLDR: A lot of money is being spent on AI in circumstances closely resembling the dot-com bubble. Startups are getting bought out quickly, and big chunks of money are moving in circles. When and how it will pop, no one knows, but Gen Z beware.

Articles:

I would be remiss to move onto other things without first talking a bit about the word that’s been everywhere the last few months: agents. If you need the lowdown on what exactly an agent is, my fellow Stack Overflow writer Eira May (hi Eira!) wrote about it at length during the original agent boom in early 2025. The quick definition of agents are AIs that are focused on making decisions, not generating content. They operate like very advanced automation systems, able to work through a long list of complicated tasks towards a specified end goal for the user. And, you know, what’s cool about agents is they’re…just little guys, you know? They’re little helpers, trying to make your life easier. And ultimately they’ve beat out GenAI as the buzzword of the year, and for good reason (Ryan wrote an excellent blog on why, which you should check out).

One thing that’s risen up around agents since they became the most talked-about form of AI is whether they can do what they say they’re gonna do. Also, a lot of people are wondering if people are actually going to use them. You could throw a stone at our podcast and hit some really cool company doing some really cool thing with agents, but at the end of the day the question still stands: How commonplace will these tools become as AI reaches maturity?

We’re starting to reach an inflection point where a lot of this tech is not moving as fast as we thought it would, and not yet doing the things we were promised it would do. And why, you may ask? If you were paying attention before, you can say it with me now…BECAUSE OF THE AI BUBBLE.

All the bullish spending led to outsized hype, which led to more spending, which led to more hype, and on and on it goes, until it pops. A lot of things were promised early on with AI, and now that the technology is reaching some level of maturity, the roadblocks to those promises are starting to become visible. For instance, it has yet to steal my job, and everyone said it would. Probably because it gives people the ick to read AI slop.

One of the issues with agents—LLMs as a whole really—is that they’re non-deterministic by nature. It’s what makes them so powerful, but also a little sloppy. The more developers use AI, the less they trust it, which is a problem if you’re a company that wants people to use AI. For a while now, people have been trying to figure out how to place guardrails on AI, and now agents specifically. Without them, LLMs’ non-determinism makes its outputs squishy and unreliable, leading to inconsistencies and mistakes that could be dangerous. We saw this happen in August when the Tea App, which was allegedly vibe coded, was hacked and the sensitive information about its all-women user base was leaked. I have not even begun to mention guardrails to keep agents from I Have No Mouth, and I Must Scream-ing us, which I think about constantly. Anyways.

Recently, the work enterprises are doing around agentic AI has become slower, but it's also more focused. This was something I saw and wrote about at Microsoft Ignite. Companies are starting to stay in their own lanes and work on tooling that is actually helpful to their customers, as opposed to just creating something sparkly that no one really wants.

Still, they want us to believe everything is going exactly to plan. Ryan experienced a flurry of agentic AI announcements while he was at re:Invent this year, but when he sat down with one of the (snarky) experts, there was definitely a little pushback on the excitement. Some are even saying AI is just the normal next evolution of technology and not something to be excited about.

Even if there are still questions around agentic AI, knowing about them is important because you’ve probably experienced them in your daily life, whether it be a vibe-coding bot or yelling, “Talk to a representative! Talk to a representative!” into the phone. Someone is yelling that somewhere in the world, right now, as we speak.

TLDR; Agents have taken over as the predominant form for AI, meaning they’re everywhere in tech right now. Whether they will actually be widely used is still up for debate, partially due to LLMs’ non-determinism, which makes agents unreliable and inconsistent. The next big step for companies is to start addressing this with guardrails.

Articles:

It really amazes me that one of the terms that dominated the tech landscape this year—vibe coding—originated from a Tweet posted a mere ten months ago. Whatever, call me Unc, I know this stuff happens all the time (it’s why In n Out isn’t using the number 67 anymore), but it still goes to show how hyped AI has been in the last year, and how quickly it has proliferated in our culture.

For those who live under a rock (and if you do, let me know if the rent is cheap because I could get into that), vibe coding refers to pretty much any type of coding that uses AI to generate code. (Ed. note: Well, sort of. It's the autonomous generation of it, where you can create a whole app without looking at code. The legit version is agentic AI code generation, but the OG version two years ago was CoPilot and copy and paste from ChatGPT.) This could mean creating pretty much all the code for an app, like I did when I vibe coded a toilet app, or smaller swathes of AI-assisted coding with functionality similar to autofill. The definitions of “use” for a lot of these new technologies are fuzzy, partially because we’re all still pretty fuzzy about how we’re going to use them.

This is by no means an exhaustive list, but here are a few ways vibe coding has rocked the tech world:

It’s given nontechnical people the tools to create apps and software, with varying degrees of success (failure: Tea App data breach vs. success: my really cool toilet app).
It’s created quite an existential crisis within the software development community about how, when, and why to implement these tools into their work.
Its existence has made getting a job as a junior developer really hard.

I’ll get to that last one later, but let’s talk about number two (toilet app pun) real quick.

Like any new tech, there are skeptics and optimists about this particular subset of AI. Developers sit on both sides—there are some that are really excited about vibe coding, some that absolutely hate it, and many that are somewhere between those two. That being said, there’s been quite a lot of hullabaloo about how agent-assisted coding is supposed to make developers 10x more productive. Which, like, maybe? The 10x developer mythos has been around forever, but if it was ever going to be achieved, it would probably be now, when we have all these super-powered coding tools.

These tools really are powerful. I found this out myself when coding my aforementioned silly toilet app for the blog. I was able to create a functioning app with zero prior coding experience, all through a vibe coding bot. My app was riddled with dangerous security holes, but in the hands of an experienced developer, these tools do have the potential to really up one’s game.

However, allow me to harken back to a time long ago, when you were reading the previous section of this blog. Remember how I talked about that non-determinism issue? Well, the non-determinism of LLMs has made vibe coding ticky for a lot of folks. Replit’s tool was recently in hot water for deleting an entire database after being explicitly told not to. Fast Company reported that we’re having a “vibe coding hangover” because developers are starting to wonder if it takes more time to fix poorly generated code than it would take to just do it themselves. Our very own Eira May (hi, Eira!) wrote about how AI coding assistants are making imposter syndrome worse for developers and that the over reliance on coding bots causes skill atrophy.

What’s become apparent is that vibe coding cannot be solely relied upon for writing code bases. Developers are being asked to shift how they’re thinking about their work to include less manual coding and more strategy and architecture. Some are even saying vibe coding requires more critical thinking for developers, not less.

All this change to the time-honored traditions of software development is having ripple effects across tech as a whole, and it means that the work of young developers—and young people in tech, overall—is due for massive shifts in the coming years. That is, if Zoomers can get a job at all…

Now’s a good time to “girl, whatever” this, because we need to talk about the job market.

TLDR; AI coding assistants and code generation apps have taken the software development world by storm. While some are saying vibe coding can make developers 10x more productive, others believe we’re actually wasting more time with poorly generated code.

Articles:

If you’re anywhere near my age, this is probably the thing you care most about: whether AI is taking all of our jobs or not. I’ve written at length about this because, based on the statistics, the answer is yes—sadly, at the time of my original report, entry-level tech hiring had dropped 25%.

Look, I am far from a “technology is bad fire is scary thomas edison was a witch” type of person, but the numbers aren’t great. Where Gen Z is right now is a frankly very weird place to be. Many of us are just entering the workforce—post-pandemic, mind you—only to find that the skills we were taught in school are now obsolete. These schools—the ones where we received expensive degrees that were supposed to give us the security of a white collar lifestyle—aren’t teaching AI skills, which are what many young tech workers need to get hired. And even if they were, the dudes who are making the AI tools say we aren’t going to get hired because there will be no need for entry-level employees.

I am clearly very chalant about this…not an ounce of nonchalant in me right now. The struggle is real. And there hasn’t been much in the way of solutions for Zoomers. We’re already worried about brain rot and climate change and school-based violence, and now we have to worry about being broke and unemployed, too. As The Washington Post put it—in a rather how do you do, my fellow kids way—“In the job market, Gen Z is cooked.”

But all is not lost. Ultimately, without junior developers there will never be any senior developers, so let’s hope they come to their senses before we’re all old and grey. And there are still many ways that junior-level devs—and juniors in any industry—can stay competitive. One is by leveling up your AI skills (ironic, I know).

We had Secure Code Warrior’s CTO Matias Madou on the pod recently, and he said that junior developers seem to have a natural speed and flexibility when it comes to adopting and learning new AI tools, giving them an edge over more senior teammates who may be stuck in their ways. Tom Moor, head of engineering at Linear, said much the same about the junior developers he had hired: Gen Z is the best at augmenting their work with AI tools. Okay! Don’t hurt ’em now, junior devs!

There’s no real answer to the conundrum that Gen Z faces with the job market, but getting really good at the latest tools and technologies may just give you the little boost you need to bag that job. Keep your enemies closer and all that. If you’re in need of it, we’ve got a ton of resources on getting good at this kind of stuff using Stack Overflow.

TLDR; The job market is not easy for junior developers—or any junior-level employee, really. While there are no easy solutions just yet for Zoomers, leaders are starting to notice that when Gen Z leans into AI tool use, they are much better at augmenting their work than their more senior counterparts. Sharpening your AI skills can be one way to give yourself an edge in this dire job market.

Articles:

Google did their big one with Quantum Computing. They released their experimental Willow chip with error correction. The more qubits used, the more errors are reduced, addressing one of the biggest challenges of quantum computing. And I was like, “Right, right, right, exactly…” I mean, I would have told them to do that if they had asked me.
Who put this watermark on my AI? Oh, it was the AI. OpenAI, Adobe, and Microsoft have implemented C2PA standards across their AI-generated images, which includes a watermark stating it’s AI. The hope is this will help us identify deepfakes and AI-generated content more easily, although you can apparently jailbreak it pretty quickly by running it through Photoshop.
Don't talk to me or my humanoid robot factory-worker son ever again. Tesla is ramping up to put a ton of humanoid robots into their factories. The humanoid part is the differentiator between these robots and the industrial ones they already have at factories, meaning it can perform a wider range of tasks. Also, if you have $13k, you can get yourself a Unitree G1, because why not? Honestly, I feel so maternal towards him.
Yeehaw, the chips are here! They’re building a $500 billion AI infrastructure in Abilene, Texas called Stargate. By they, I mean basically everyone you can think of—SoftBank, OpenAI, Oracle, MGX, Arm, Microsoft, and NVIDIA. The first phase is being powered by 50,000 NVIDIA Blackwell chips because chip shortage who? They—the collective, powerful they—hope to circumvent some of the computing power ceilings they have been facing with this project, allowing them to scale as they need in the privacy of their own $500 billion Texas town.
Sorry, can’t work, there’s a cloud outage. There’s been a surprising amount of cloud outages this year, with AWS, Google Cloud, and Microsoft Azure all going down, and back-to-back Cloudflare outages just a few months ago. AI’s demand for compute and energy are the most likely cause for these, and while the Texan born-and-bred Stargate should take some of the pressure off, there’s still the question of scaling cloud fast enough to keep up with AI’s need.

And there you have it! There’s the lowdown on everything you, dear Zoomer reader, need to know about the 2025 tech landscape. And with that, the 2025 season comes to an end. And with the dawn just now breaking on 2026, there’s surely much more to come—like GTA 6 or 7 maybe—but until then, I’ll see you on Stack Overflow.

If you're a Zoomer, this one's for you: Everything Gen Z needs to know about the 2025 tech landscape

The concept of an AI bubble...

Agents, iykyk

The vibes are…in shambles?

God forbid a Zoomer wants a job in 2025

And here’s a few other things you should know from 2025:

Add to the discussion