Software Developer

Software Developer scores 17/100 on this site's structural exposure measure, where lower means more exposed (Medium confidence, two independent pillars). Both the OpenAI "GPTs are GPTs" task ratings and the AIOE index read this work as highly exposed, which is unsurprising: the output of the job is text (code, tests, documentation, tickets) produced at a keyboard, and that is the kind of work language models are built to produce.

The score describes tasks, not headcount, and it cannot tell you whether AI will replace the occupation, so the rest of this section covers what is actually happening in the occupation.

Use of AI tools is already mainstream. In the Stack Overflow 2025 Developer Survey, 50. 6% of professional developers said they use AI tools daily and another 17. 4% weekly, and 30. 9% of all respondents use AI agents at work. Anthropic's Economic Index report of March 2026 (conversations sampled 5 to 12 February 2026) found that Computer and Mathematical tasks made up 35% of Claude. ai conversations and that coding remained the most common use.

Nobody has to guess whether developers are using these tools. They are.

The evidence on how much the tools help is more mixed than the adoption numbers suggest. In the same Stack Overflow survey, 46% of respondents said they distrust the accuracy of AI output, 66% named "almost right, but not quite" as their top frustration, and 45. 2% said debugging AI-generated code takes more time.

A 2025 randomized trial by METR gave 16 experienced open-source developers 246 real issues; with AI tools allowed, they took 19% longer, while believing they had been about 20% faster. METR itself calls that result out of date for current tools and says it does not show AI fails to help most developers, so it is a caution against assuming large gains, not proof of small ones.

Hiring is where the pressure shows up. Salesforce CEO Marc Benioff said on the company's May 2026 earnings call, as reported by Fortune, that "we're not hiring more engineers" and that engineer headcount had been "mostly flat" for the last couple of years because the company uses AI to make its engineers more efficient. We could not read a primary transcript of the call.

A Stanford Digital Economy Lab update (August 2026, ADP payroll data through June 2026) found employment of 22- to 25-year-olds in highly AI-exposed occupations about 19% below its counterfactual, with no comparable gap for experienced workers; the authors stress these are descriptive patterns, not causal estimates, and that figure covers all highly exposed occupations, not developers alone. Indeed Hiring Lab reported in July 2026 that US software development postings were still about 27.

5% below February 2020 but had risen almost 15% since February 2025, and that 71% of the increase from May 2025 to May 2026 came from senior roles. A later reading of the same Indeed index (published on FRED as series IHLIDXUSTPSOFTDEVE) put postings about 21% below February 2020 as of 16 October 2026. Taken together, these point to a squeeze on entry-level and junior hiring more than a collapse in total demand.

The counter-evidence is real and it is why this page does not predict mass job loss. BLS projects employment for software developers, QA analysts, and testers to grow 10% from 2025 to 2035, adding 185,400 jobs, with about 106,100 openings a year, against 3% for all occupations; software developers alone are also projected to grow 10%, adding 174,700 jobs.

BLS's own reason is that demand "is projected to be strong due to the continued expansion of software development for artificial intelligence (AI), Internet of Things (IoT), robotics, and other automation applications. " Pay is high: the BLS May 2025 median for software developers is $135,980, with the 10th percentile at $82,460 and the 90th at $214,670.

The 2025 BLS Monthly Labor Review analysis of the earlier 2023-33 projections concluded that software developers were unlikely to see an employment decline, noting that AI suits the occupation's tasks but could also raise demand for software, and called the effects highly uncertain. Those are projections, and a 2035 projection says little about who gets hired in 2027.

What the checklist finds missing is any structural wall. BLS names a bachelor's degree, not a license, as the entry requirement, so no credential is tied to the person doing the task. The work is digital, so physical-presence protection does not apply either. Real accountability exists in practice, since a human engineer owns the code that ships and the pager that goes off, but under this site's research checklist that corroborates a protective case and cannot build one alone.

Applying the rubric as written, high exposure with no license wall and no physical wall gives At risk. Here that label means exposed with no structural protection, not that BLS projects fewer developers.

For someone choosing or staying in this career, our reading (an inference, not a sourced finding) is that the part of the job most likely to shrink is turning a clear specification into routine code, and the part most likely to hold value is deciding what to build, reviewing machine-written changes, and owning systems in production.

The next-best option for people who like building but want a physical-world anchor is a hands-on technical trade where programming knowledge still helps, such as the robotics and industrial electronics roles listed below. Those pay less on this site's figures, a real trade-off.