Every few days, someone confidently declares that AI is about to wipe out software engineering.
I don't buy that.
I think software is much closer to woodworking than people realize.
Before industrialization, furniture was made by hand. If you wanted a table, a chair, a cabinet, or a bed frame, you needed a skilled craftsperson. The work took years of training. It was slow, specialized, and expensive. Good furniture was not broadly accessible because it could not be. Every piece required real human expertise, and that expertise did not scale cheaply.
Then industrialization happened.
The craft of making furniture did not disappear. It changed.
Machines, standardization, and repeatable manufacturing processes made it possible to produce furniture at much larger scale and much lower cost. The average quality was often lower than what a master craftsperson could produce by hand, but that was only part of the story. The bigger change was that an entirely new market opened up. Furniture became accessible to far more people. Businesses like IKEA could exist. A thing that once required expert labor for every unit could now be produced in enormous volume by a system.
Handmade furniture did not go away. There is still a market for beautiful, custom, ultra-high-quality work. But the center of gravity changed. Some artisans were displaced. Some adapted. Some moved further upmarket. And alongside all of that, a much larger market for cheaper, faster, more standardized output came into existence.
I think software is going to follow a very similar path.
Software Has Been Craft Work for a Long Time
For most of its history, building software has looked a lot like skilled handcraft.
You needed people with deep, specific knowledge. You needed years of training. You needed teams of specialists who could translate ambiguous human needs into precise technical systems. Every meaningful piece of software required a large amount of manual effort from people who knew what they were doing.
That made software powerful, but it also made it expensive. There are countless products, automations, internal tools, and business ideas that never got built, not because they were bad ideas, but because the cost of making software was too high. The custom work involved was too specialized. The economics did not work.
AI Lowers the Cost of Production
AI is making it cheaper and faster to produce software that is good enough for many use cases. Not perfect. Not elegant. Not deeply original. But good enough.
There are companies that will now build internal tools they never would have funded before. There are small businesses that will buy software they could never previously afford. There are workflows that stayed manual for years because no one was going to assign a full engineering team to automate them. There are product ideas that used to die in a spreadsheet that may now become viable.
The important question is not just whether AI can replace some portion of current engineering labor. The important question is what happens when software itself becomes far more manufacturable.
When that happens, you don't just get substitution. You get expansion.
You get more software.
And Yes, a Lot of It Will Be the IKEA Version
Good enough code that is customizable enough for most applications.
When you industrialize a craft, you get standardization. You get lower costs. You get accessibility. You also get a lot more average output.
Some of it will be flimsy.
Some of it will be ugly.
Some of it will work surprisingly well for the price.
Some of it will be disposable.
Some of it will be “good enough” in exactly the way the buyer needed.
There will absolutely be a flood of mediocre AI-assisted software. There will be overconfident companies, sloppy implementations, badly framed cost-cutting strategies, and executives who mistake faster production for complete replacement of judgment. There will be chaos, overcorrection, and a lot of noise.
Some people will be displaced. Some categories of work will shrink. Some tasks that used to justify a full human workflow will get compressed into something smaller.
That is not the same thing as saying software engineering jobs disappear. It means the structure of the market changes.
Industrialization Changes the Work Before It Eliminates It
People talk as if the only possibilities are: either software engineering remains exactly what it has been, or it vanishes, but that is usually not what happens to crafts or professions.
The tools change first.
Then the workflow changes.
Then the expectations change.
Then the skill mix changes.
Eventually, the job title survives while the actual work looks very different.
Software engineers may spend less time manually producing obvious code and more time shaping systems. Less time writing boilerplate and more time defining constraints. Less time assembling known patterns by hand and more time evaluating tradeoffs, validating outputs, debugging weird behavior, understanding real user needs, and deciding what quality means in context.
That is still engineering. In some ways, it is more engineering and less typing.
High-Craft Software Will Still Exist
Even after industrialization, handmade furniture did not become pointless, it became differentiated.
If you want something deeply customized, unusually durable, aesthetically excellent, or built around exact needs, you still want real craftsmanship. In fact, the abundance of cheap mass-produced alternatives can make true quality stand out more, not less.
I think the same thing will happen in software.
There will still be demand for systems that are reliable, maintainable, secure, deeply integrated, and thoughtfully designed. There will still be software where correctness matters. There will still be edge cases, constraints, and human realities that don't fit cleanly into standardized generation.
When code becomes easier to generate, discernment becomes more valuable. When anyone can produce output, the differentiator shifts toward knowing what should be built, how it should be shaped, where the risks are, what corners cannot be cut, and what makes something actually good.
The Bigger Story Is Market Expansion
One reason I am skeptical of the “all jobs go away” framing is that it assumes the size of the software market stays basically fixed. I don't think it will.
When making something gets cheaper, people don't merely buy the old amount for less money. They start buying things that previously were not worth buying at all.
Cheap furniture did not just replace expensive furniture. It furnished more homes.
AI-generated and AI-assisted software will not just compete with existing software teams. It will also create a much larger market for software in places where custom development was previously too expensive, too slow, or too inaccessible.
Some of the work will move downward into highly standardized production. Some of it will move upward into higher-leverage oversight, architecture, integration, and quality. Some entirely new categories of work will appear because the economics now support them.
That seems much more plausible to me than “software engineering stops existing.”
I’ve Heard This Story Before
My dad worked in television for about forty years. He started in the late 1970s, right as one era was ending and another was beginning. The first great golden age of television was fading, and cable was rising into what you could probably call a second golden age.
For a long time, the industry changed less than people might expect. The scale grew. Availability exploded. More channels, more content, more reach. But the basic operating model and economics stayed recognizable. Television was still television. Prime time still mattered. The machinery of the industry still made sense to the people inside it.
Then the 2010s happened.
Streaming did not just compete with network and cable television. It reorganized the entire landscape. Very quickly, the old structure stopped being the center of gravity. The audience changed. The business model changed. Distribution changed. Viewing habits changed. Many of the jobs my dad did over the course of his career simply do not exist anymore.
The industry did not die, it transformed into something that most people inside it probably could not have fully envisioned until after it happened.
Johnny Carson was eventually followed by Letterman, Leno, Conan, and then a parade of Jimmys. But even that version of the industry turned out not to be permanent. It was gradually overtaken by something far more fragmented: MrBeast, the Paul brothers, YouTubers, Twitch streamers, podcasters, TikTok creators, and countless internet-native personalities who built audiences outside the old system entirely. The old model was not replaced by a new host. It was replaced by a new landscape.
My dad jokes that prime time is what paid for his retirement, and I think that line captures a lot. That whole world was real. It was massive. It shaped careers, businesses, and lives. It also largely does not exist anymore in the form that once felt permanent.
That does not mean television ended. There is more video content than ever. More ways to watch it. More creators. More distribution. More total output. But the old structure was not preserved just because it had once seemed foundational. It was replaced by a new one.
I think software may be heading for a similar kind of shift.
For People Who Love Solving Problems, This Could Be a Golden Age
A lot of programmers don't actually love typing code for its own sake, they love solving problems.
They love understanding systems. They love building tools. They love making messy things clean. They love turning ambiguity into structure. They love the feeling of gaining leverage over a problem that looked impossible a few hours ago. That part is not going away.
The problems are changing. The tools are changing. The constraints are changing. The surface area of what one person can build is changing. For people who do this for the love of the game, that is exciting because everything will be in motion.
There will be new abstractions, new workflows, new product categories, new expectations, and new kinds of technical judgment. There will be new opportunities to build things that were previously out of reach. There will be a massive amount of room for experimentation.
If you like problem solving more than you like protecting old workflows, this moment is full of possibilities.
The Transition Will Be Messy
Industrialization is disruptive.
There will be companies that overcorrect badly. There will be teams that are reduced too aggressively. There will be low-quality software shipped at scale because someone decided speed mattered more than understanding. There will be a period where many people are asked to operate in systems that have changed faster than the norms around them.
If your current job is mostly composed of work that is easy to standardize, I don't think denial is a useful response. The market is changing. Pretending otherwise will not help. The better response, at least to me, is adaptation.
Learn the tools.
Understand the new economics.
Move toward the parts of the work where judgment matters.
Get better at problem framing, system design, evaluation, communication, and taste.
Become the person who can work with the new machinery without mistaking it for magic.
That seems like a far more grounded response than mourning a version of the craft that was never going to stay frozen in time.
Some jobs will shrink. Some will disappear. Some will be reborn in altered forms. Entire new layers of software work will appear. New tools will create new expectations. New economics will produce software that would never have existed under the old model. And a lot of what feels central today may someday feel like prime time television: once dominant, deeply real, and eventually no longer the thing that anchors the industry.
If your goal is to keep building, keep learning, and keep solving interesting problems, it might be the most interesting moment the field has seen in a very long time.
The craft of making software is not disappearing. But like every craft touched by new machinery, new economics, and new distribution, it is going to change shape.
Some people will spend too much energy mourning that.
I would rather spend mine learning the new tools and exploring what becomes possible next.
Top comments (0)