IT engineer pay polarises as AI takes hold

2026-07-19

Gap widens between upstream and downstream work

The spread of job substitution by artificial intelligence is beginning to show up in pay gaps for IT engineers. Since 2022, rates for work that is difficult to replace with AI have risen by 20%, while downstream work, which is easier to automate, has fallen by 50%. Employment shrinkage that has already taken hold overseas is now starting in Japan, leaving workers to rebuild their skills.

Nikkei analysed about 110,000 compensation rates offered by companies to freelance IT engineers, with the cooperation of Levtech, an IT talent services company based in Tokyo's Shibuya district. Job trends serve as a leading indicator of corporate hiring trends. The figures were indexed to 100 in April 2022, before the debut of OpenAI's chatbot ChatGPT.

Code generation drives substitution

IT engineers' work is broadly divided into 'upstream work', which involves organising customer requirements and handling basic design, and 'downstream work', which covers coding and testing. The rate for project managers, a typical upstream role, rose by 20% in May 2026, while the rate for HTML coders, a typical downstream role, fell by 50%.

The gap has become even more pronounced since 2025. The timing coincides with the spread of 'vibe coding', in which most code creation is left to generative AI. At Levtech, the number of intermediary placements in upstream work rose more than fivefold, while downstream work, which is more vulnerable to AI, fell by about 30%.

AI tools specialised in code generation, including Anthropic's Claude Code, have improved dramatically in accuracy and are accelerating labour substitution, said Narinori Ashino, a recruiting adviser at Levtech. From 2026 onward, more companies have begun to stop hiring engineers who handle coding.

Upstream work, which involves coordination and negotiations with stakeholders, is difficult to replace with AI in the short term. A new role called 'forward deployed engineer' or FDE, which is stationed at client companies to support AI adoption, is also spreading.

The IT sector still faces labour shortages, and demand for engineers remains firm. Ashino said hiring appetite is rising for people who can use AI effectively while delivering better results.

Differences also seen in duties and promotion

Whether workers can use AI effectively will shape future career development. In a survey Nikkei conducted with Levtech among 572 IT engineers in their 20s to 50s, 65% said income gaps will widen in the future depending on AI skills.

Asked about AI-related gaps already emerging at work, 51.3% cited 'the scope of work they can handle' as the biggest factor. 'Evaluation and promotion speed' was next at 39.5%.

At Fenrir, an app development company in Osaka's Kita ward, the amount of code generated by about 20 engineers using the same AI tool differed by as much as 700 times in autumn 2025. The case showed how productivity can vary greatly depending on how AI is used and integrated into work.

A 41-year-old male employee who demonstrated high productivity through effective use of AI was later appointed to lead a key in-house project. Kentaro Maegakiuchi, head of the cloud-native technology department, said AI proficiency will become an essential skill for IT engineers.

In a 2025 report, the International Labour Organization identified software development as one of the occupations heavily affected by generative AI. The changes taking place among IT engineers can be seen as a leading example of the labour market in the AI era. More broadly across white-collar work, the ability to use AI is increasingly determining the value of talent.

Enjoyed this article? Share it with your network!