TrueUp is tracking over 67,000 open software engineering roles globally, the highest in more than three years. The headline is correct. What the number does not tell you is what those jobs actually require now.
TrueUp's hiring data shows over 67,000 open software engineering positions globally, the highest it has been in more than three years. The engineers struggling to fill them are the ones still waiting for the job to go back to what it was in 2022.
I have heard the "AI will kill software jobs" take nonstop for two years. The numbers still do not back it up.
TrueUp, a hiring analytics firm that tracks real-time tech job postings, is currently tracking over 67,000 open software engineering roles globally. That is a 30% increase since the start of 2026 and roughly double the number of openings from the market low in mid-2023. TrueUp founder Amit Taylor has been direct. The data does not support the claim that AI is replacing engineers.
The easy headline is "see, everything is fine." It is not that simple. The jobs are there and the bar moved.
Companies are not posting 67,000 openings to rebuild 2022 teams. They are hiring for a different profile.
You can see it in team structure. Companies that cut in 2023 and 2024 did not rebuild the same org chart when hiring returned. They rebuilt leaner. Fewer people. Wider scope. Higher expectations per engineer.
In my experience, teams now want people who can use AI speed without losing judgment. That part matters more than raw output.
The TrueUp chart should make you ask one practical question. Who owns all the extra code now getting shipped faster than ever.
AI tools can generate whole features in a fraction of the old timeline. Output goes up. Ownership does not disappear.
Amazon learned this in a very visible way. An AI-assisted deployment caused an outage because the review layer that would normally catch the issue was not built to handle the speed at which AI was generating and pushing code. The AI moved faster than the human oversight could keep up with.
That incident was not about a careless model. It was a process problem. The review system was built for human pace, then asked to govern machine pace.
More open positions means more surface area. More generated code means more places where someone needs to understand what shipped, even if they did not type every line. Responsibility never transferred to the model.
Teams are still hiring people who can execute tickets. They are not paying a premium for engineers whose whole game is ticket execution. AI compresses that loop too well.
The engineers getting hired are the ones who can:
None of these skills are new. What changed is the pricing. They used to be differentiators. Now they are table stakes.
A lot of this conversation assumes experienced engineers suddenly lost value. I do not buy that.
The roles getting squeezed are the ones built around atomic task throughput. Boilerplate production. Fast framework implementation without deep context. Clean execution of narrow requirements.
AI is very good at that tier. Engineers who built their identity around that speed are in the toughest part of the market.
Engineers who see the whole system, push back on bad requirements, and still ship are more valuable now than they were before.
The 67,000 number is real. Software hiring is stronger than the doom narrative suggests. AI did not erase the profession. It raised the floor.
Are you building the skills that make AI useful in production, or competing with it on the exact tasks it is already better at? Understand what you are building and why. Own what ships no matter who wrote the first draft. Learn to orchestrate multiple tools toward one production outcome. That is the job now.