I’m hearing it everywhere—and experiencing it myself. Software Engineers are experiencing a second youth. Even the most skeptical started embracing LLMs in the past few weeks. But the more I use these tools, the more I realize they’ve exposed what’s always been true—writing code was never the hardest part.
This inflection point raises so many questions. Is this the last generation of Software Engineers that wrote code by hand? How will future engineers develop the intuitions that you gain by building large systems by hand? Are those intuitions lost forever? And how will it shape the industry’s future? Perhaps questions that deserve another blog post.
I’ve been wondering about the increase in productivity, which, at the macro level, doesn’t seem particularly dramatic. And at the micro level, I now implement in 1 hour what used to take a day—and yet the project still ships at roughly the same speed. Perhaps it’s too early to measure, since adoption is likely very uneven and most of us are not fully realizing LLMs’ true potential. There are at least two more explanations. One is that this shift made it more apparent that implementation time is not the bottleneck. The real bottleneck is figuring out what to build. Another explanation is that Amdahl’s law is at play. We’ve improved code generation time, but that’s just a fraction of the development process. I find myself spending less time on coding but more time on scoping, project definition, and QA.
Regardless of the actual productivity improvements, I fear that market pressures, combined with a limited understanding of current model capabilities, might lead to premature task force reductions—more layoffs, in human speak.
Then there are all the posts invoking Jevons Paradox with the promise of increased demand for software. I’m not convinced it’s that simple. If it’s true that companies can continue to grow with just 30% of their current software engineers, what are they going to do with the remaining 70%? Will large companies expand into new verticals? Perhaps. But large organizations are not known for moving at the pace of technological change. If companies decide to just layoff their surplus engineers, will there be a boom in startup creation? I don’t see that happening either, when the bottleneck is deciding what to build. And if there were a start-up boom, competition would be fierce, as good ideas are scarce. If the scarce resource is knowing what to build, making code cheaper won’t unlock much new demand.
My recommendation? Embrace change, get used to rapid adaptation, and get good at judging what is worth building in the first place. Lastly, brace yourselves: 2026 will be a hell of a roller coaster.
