In 1865 an English economist named William Stanley Jevons noticed something odd. Steam engines had become more efficient and burned less coal per unit of work, so everyone expected Britain to use less coal. The opposite happened. Once steam power got cheap, it went into factories, railways and mines that couldn't afford it before, and coal use jumped.
Economists still call it the Jevons paradox, and it's the best answer we've found to the question every manager is asking right now: if AI writes code, why do we still need developers in the AI era?
When building gets cheaper, people build more
AI has lowered the cost of writing code. That's true, and we see it every day. Now look at it from the customer's side. The internal system a mid-sized business gave up on three years ago because it cost too much suddenly fits the budget. The automation that was a nice-to-have becomes a quarterly project. Every idea stuck on the "someday" list now gets a quote.
And every one of those new systems has to be connected to what's already there, secured, maintained and updated as the business changes. More software in the world means more work for people who understand software.
To be fair, economists disagree on whether the paradox holds for the number of developers or only for the amount of software. It depends on how elastic demand for software really is. So let's look at what the market actually shows.
What the data shows
The US Bureau of Labor Statistics, which publishes the official employment projections, expects jobs for software developers, QA analysts and testers to grow 10% between 2025 and 2035, with about 106,100 openings a year on average. Even with AI tools everywhere, the official forecast points up.
The more interesting picture comes from Stanford. Researchers at the Digital Economy Lab looked at payroll data from ADP, one of the largest US payroll providers. In jobs where AI is most embedded, employment of 22- to 25-year-olds fell about 13% relative to the rest of the market. For experienced workers in the very same jobs, employment held steady or grew. The researchers are careful here: they say AI doesn't explain the whole decline, and high interest rates played a part.
So the market still needs developers. It's mostly looking for experienced ones.
What actually changed
A lot of the work junior developers used to do, writing simple code from clear instructions, is exactly what AI does well today. What's left, and matters more now, is the part that takes experience:
- Deciding what to build. Turning a business need into a system, and knowing which feature to drop because it will cost a fortune to maintain.
- Connecting to the real world. A new system doesn't live alone. It has to talk to the CRM, the accounting system and ten-year-old data.
- Checking what the AI wrote. Code that works in a demo still has to prove it can handle a thousand users on a Monday morning. People who have built systems like that spot the difference fast.
- Maintaining it for years. A large share of the work on software comes after launch, when it needs changing, fixing and extending.
Gartner reached a similar conclusion. It forecasts that through 2027, about 80% of the engineering workforce will need to upskill because of generative AI. Its analyst Philip Walsh put it this way: AI will transform the role of software engineers, but human expertise and creativity will remain essential for delivering complex, innovative software.
The developer the market wants today runs AI like a work tool and takes responsibility for the result.
What to ask a developer or a team in 2026
If you're looking for development help now, here's what we would ask:
- How do you use AI in your work? A good answer names specific tools and says where they don't trust them. Someone who doesn't use AI at all will be slow. Someone who can't explain how they check what the AI wrote will hand that checking to you.
- What have you built that's similar to what we need? Relevant experience is worth more now than ever, because it's the one thing AI doesn't bring.
- Who owns the big picture? Someone has to see the whole system, not just this week's task.
- What happens after launch? Who maintains it, who documents it, and confirm the code belongs to you.
Where we come in
Avense Tech operates in Israel and the Philippines, and we offer three services: a development team that joins yours, a full project from spec to launch, and management of your development team. In each of them, AI is part of the toolkit, and responsibility for the result stays with people.
The bottom line
AI made building cheaper, so businesses are building more software. What changed is who's in demand: fewer hands typing, more experience and judgment.
Want to talk about what you need to build? Talk to us.





