The ouroboros industry
An honest look from within American software development.
January 1 marked four years since I professionally entered the software industry, all largely spent as a data engineer. Compared to some of my mentors and team leads, most of whom have decades of software development experience, the four-year mark is relatively speaking the blink of an eye. Yet in that time, I've experienced at least one CEO change and one layoff every year — emphasis on at least one, some years brought doubles. I've played key roles in multiple projects which individuals multiple pay grades over me have described as "once in a lifetime".
Of course none of this is a dig at my employment, I've met some incredible people in this industry: people who took the time to answer questions, hand down their experience, and collaborate with grace. (To these individuals, if you're reading: my sincerest thanks.) All this talk of layoffs and regime change is purely to lend some credence to some of my thoughts on this line of work after the world ended in 2020, professionally the only world I've ever known.
A picture of software
Since 2022, the American tech sector has laid off over half a million headcount. With estimates saying there are roughly 16.1 million jobs across all tech, napkin math says 1 out of every 50 American tech workers have been laid off in the past 4 years. Many of these positions are dissolved entirely or shipped out-of-country due to some nice tax incentives. This doesn't even take into account the AI tailchasing that has been responsible for nearly 72,000 of those job losses in tech alone.
Now taking a lateral step from the numbers, the feeling on the ground is a pretty bleak one.
Break glass in case of mild anxiety
I've seen layoffs — supposedly an emergency measure before my time — become a household tool for executive leadership seeking a quick attaboy from the board. One can only hear "these measures aren't taken lightly" so many times, be promised "no layoffs past this point" before growing jaded.
Mind that these very same corporations are seeing record profits year over year and the market has never been higher. The ripcord is pulled even when the company isn't embroiled in mortal struggle. I've personally seen layoffs invoked when a growth target is missed by single-digits; the company still grew, was still funded, and yet the news of health brings the same reward as if the company were minutes from breathing its last.
It's us on the ground that are axed for a missed target rather than the board simply asking themselves if their own expectations were faulty.
"Leaders", rapport, and the problem of trust
Outside of legacy Silicon Valley unicorns, CEOs have entered their revolving door era, passed between companies like consultants rather than captains. They don't carry the burden of building trust; if you're lucky they'll stick around for maybe two years before either their performance fails to entertain the shareholders, or they dive for the exit. The time between announcement and transition is shrinking, as well as the time it takes for a new CEO to replace downstream leadership positions with their preferred crew.
These takeovers strip away one of the few powers left at the individual contributor level: rapport. Establishing good working relationships with colleagues, managers, and directors is a hedge play and a huge part of why layoffs were supposedly so difficult in the first place. There's just so much more social friction involved when you can put a name and a face to the person you're screwing over.
But with frequent layoffs and restructuring, the cultivation of allies in the workplace can't pay dividends like it used to. All of that time spent advocating to a direct manager or skip-level or even an executive was for naught and now you have to rebuild your case from scratch. If you or anyone can't piece together a network of rapport, trust simply disintegrates. ICs have no reason to trust a new C-suite, and the new C-suite doesn't have to acknowledge previous alliances if they don't want to, and so all that remains is the imbalance of power and a swath of disconnected employees.
Couple this with a limping software industry and it makes sense that CEOs don't care to establish trust with developers. The threat of leaving for greener pastures only works if there are greener pastures to leave for, and so anyone still fortunate to find themselves employed are, for the time being, captive while also being in no way shielded from weight-shedding directives from on high. Not loyalty, not expertise, not tenure can save any IC from the purge rolls.
Loss of the tail
Even more concerning is the plane stall of hiring since 2020.
Junior-level developer positions especially are a dying breed, in the process of being chased to extinction by AI-truthers at the corporate top that in their infinite capacity to remain incurious do not understand that the difference between a vibe-coded college software project and one made without AI tooling is the person sitting at the keyboard. In the former is satisfaction in having made, in the latter, satisfaction in making. Taking a subpar codebase and continually refining, rebuilding, polishing to a mirror shine is the behavior and mentality of the craftsman. Generating a subpar codebase to then plead with agentic systems for improvement or to convince onlookers that "no, it's actually solid" is not. The cruel irony of all this is that regardless of the usage of LLMs, both have the same dismal chance at employment...because of LLMs.
Nothing makes this clearer than a report released by ADP that finds that the average American software developer is 38 years old. If averages aren't your speed, then the cohort of 21 to 25 year-old tech workers slipping from 15% of headcount in 2023 to just under 7% in 2025 — cut by more than half — is undeniable.
This same report shows consistent employment losses year-over-year since the pandemic, likely attributed to near-zero interest rates exploited by venture capital finally running out. What I've found this to mean practically is that layoffs without replacement of headcount require those who are left to generate more value with less support.
The same can be said of non-engineering jobs within the tech sector. Sales, management, HR, business administration, and analytics are all even more vulnerable to restructuring than engineers. In my experience, for every one developer cut, two managers and ten business support staff are wiped out as well.
Window-breaking machines
Growth targets aren't slowing down and expectations are blown wildly out of proportion by the promise of stratospheric productivity increases thanks to LLMs. Less time is dedicated toward tech debt and sustainment work in pursuit of revenue-positive product development. It's an unending tailchase of new, new, new without setting aside the time or even acknowledging the need to do the job well.
Where has widespread agentic tool adoption gotten us? A pretty staggering number of outages in 2025.
In a CodeRabbit article from July, Sahil Mohan Bansal wrote:
By only measuring AI use and not the quality of the code that AI generates or the actual time saved once debugging and more involved code reviews are factored in, you could be incentivizing AI use even when AI use hurts your company.
In that case, you might achieve 50% of your codebase being AI generated while also adding exponentially more bugs to your code and increasing issues and customer complaints. And to make it worse, you might ALSO not be saving any time since your devs could be spending an equal amount of time reviewing and fixing that code as they would have if they wrote it from scratch.
But your AI usage would sound impressive during an earnings call, right?
Interestingly, Bansal loses the plot a bit — after correctly blasting the myopic views of tech leadership — and attempts to come up with the middle ground of "some AI-generated code as a treat". The article fails to crescendo to the immutable fact that people are the core of the business and are widely being forced into a position of limitless consequence for the failures of agentic systems which cannot be held responsible.
Nothing left
Lately I've had siblings and cousins, in high school and approaching college, ask me if they should pursue my line of work. I can imagine that others their age might ask the same, and so this next section I want to speak directly to anyone in the US thinking about getting into this industry. I offer two answers.
The first: no. There's nothing left here for you.
Curiosity and tenacity are not valued by tech leadership at this moment. They don't care to maintain starter positions and cultivate talent. Make no mistake: they need makers, they just don't want to pay, preferring to buy into the fantasy that agentic products can replace makers with mass-theft-modeling, designed to cough up boilerplate Supabase apps with an entrepreneurcore Vercel web UI.
Elsewhere
And so I arrive at my second answer: pursue craftsmanship anyway.
Electrical engineering, woodworking, the trades, architecture, pure maths, robotics, and material science were all paths I wish I explored more in college, though they certainly aren't the only paths available. Almost all lead to industries that will always be in need of curiosity, of new talent. This in no way discounts the value of the arts, conservation, library sciences, education, and public service, all play a critical role on the world stage.
I especially want to nod to the thriving indie games industry, a field propped on pure passion that can only exist outside of the established tech sector. Moonshots are finding their audiences more and more often, many by complete accident. The same can be said of developers creating new and exciting open-source tools receiving grants from nonprofits and grateful donators alike. In these areas, there remains a place for the makers.
Regardless of the particular field, if there's an undeniable gravity to it — a fascination, a hero that you look up to, a chance to live a better life than those who came before you — then fall in. Some of these fields have well-trodden pathways from syllabus to start date. Some will be hilariously underfunded and incredibly cutthroat. In each are real challenges; all can be surmounted if they're approached with the sober facts.
Often overlooked, maybe misconstrued in a more career networking sense, is the human element. Reach out for people, listen to and understand them, cultivate a web of support since nobody goes it alone. Having good people to lean on will mean the world, just as it means the world to them when you can be leaned upon.
In software, there are no more knowns. The last thing I'd want for my siblings, or anyone else for that matter, is for them to put up 110% effort just to get legswept out of their company for the sole reason that a higher-up can only see the value in dollars saved rather than human expertise. I can only imagine how crushing it is to pursue this line of work, to have done everything right, just to one day be summarily terminated and your livelihood upheaved in one of the most uncertain moments in US history.
This state of affairs may or may not indicate the future, but it is the current reality. It is not insurmountable, but it needs to be clear that we aren't dealing in the same mythologies believed to be true since the dot-com bubble. Getting into the industry is harder and more hostile than ever before. Keeping a job in this industry is more of the same. Any reverence for the makers, the inventors, and the curious is long gone.
This is where we are
My voice is certainly not the first in all this, but I want it to be one that acknowledges that the ladder has been pulled up.
This industry is atrophying from the surface down, eating through layers of talent until it reaches the substrate of 10x engineers and legacy developers with "former Apple alum" in their LinkedIn bios, inoculated through sheer force of clout. American software development can't maintain the facade of stability, of creativity, and of opportunity anymore. It has become the consensus of those at the wheel that people are a liability, an expense on the quarterly report, and it's easier to cut down costs when there's no reason to build rapport with the ones that keep the lights on.
Employers in our sector continue to chase infinite growth — for the most part, the market has been favorable toward tech stocks — while at the same time cutting jobs yearly as if there isn't enough cash to go around. Junior positions are evaporating and the cuts are creeping up the seniority ladder. Developers are faced with the grim choice of sticking with companies that are just waiting for a reason to give them the boot or to try their luck in a sea of other employers with similar anti-maker values. Morale is at an all-time low and all that can be done is go heads-down, work harder, and be lucky during the next purge.
At the four-year mark, I've found myself frustrated by tech leadership hellbent on hollowing out the American software development workforce for short-term gain, culturally enabled by a stubborn mythology encapsulated by old videos of Steve Jobs delivering the iPhone keynote. While I'm grateful for the people I've met, the projects I've led, the sheer cash offered, I refuse to be dishonest about what I'm seeing on the inside: this industry is eating itself, and I fear soon there will be nothing left even for me.
