jobs | ai | developer jobs | it freelancer | information technology
Nepal’s IT exports have grown from about $515 million in 2022 to an estimated $1 billion, or NRs 145 billion, in 2025. While the government's service trade report recorded only NRs 22.33 billion of IT service exports in 2024/25, it acknowledges that this underestimates the sector’s actual size, as many freelancers receive payments outside the country’s banking system.
Whatever the actual figure, the ambitions riding on growing IT exports are substantial. Industry leaders estimate that around 100,000 people now work in the sector. According to the president of the Nepal Association for Software and IT Services Companies (NAS-IT), the number could reach 500,000 within a decade, 18% growth a year. That projection assumes the sector keeps hiring beginners.
I am less convinced it will. In my experience, a large share of the work Nepal sells abroad is routine coding: building pages, adding features, fixing bugs, connecting systems, etc. It’s how many fresh graduates learn the job, the kind of work AI coding tools do increasingly faster.
AI helps most with the work juniors do
In a controlled experiment by GitHub, developers using its AI assistant finished a given task 55.8% faster than those working without it. Field trials at Microsoft, Accenture and a Fortune 100 company, covering 4,867 developers, found a 26% rise in completed tasks, with the largest gains among recent hires and junior staff.
The picture changes with harder work. In a 2025 study by the research group METR, experienced developers working on large, familiar open-source projects took 19% longer to complete tasks when using AI tools, though they estimated the tools would make them 20% faster.
This suggests that AI speeds up small, well-defined tasks, but can slow down complex projects in a large codebase the developer is well familiar with. Those small tasks are exactly the kind of work a junior developer is typically assigned in the first year.
Entry-level hiring is already slowing
A Stanford study of US payroll records found that workers aged 22 to 25 in the occupations most exposed to AI have seen a 13% relative decline in employment since generative AI spread, while more experienced workers in the same occupations held steady or grew.
India shows a similar turn. Its five largest IT firms ended FY26 with 7,389 fewer employees between them, after adding 12,718 the year before. Most of the fall came from layoffs of around 12,000 people at TCS, which now plans to hire about 25,000 freshers in FY27, against 40,000 to 42,000 in earlier years.
Across South Asia, the World Bank found that job listings fell by around 20% in the occupations most exposed to AI, relative to others, after ChatGPT was released. The bank warned that young, moderately educated workers in business services and IT are particularly vulnerable.
Nepal shows similar signs. Shailendra Raj Giri, managing director of the job portal Mero Job, told The Kathmandu Post in April that IT hiring has declined since early 2024 and that demand for software developers has dropped sharply. “For fresh IT graduates, the market looks almost dark right now,” he said. “Companies have stopped hiring new people altogether.”
In the same report, Sunaina Ghimire Pandey, president of the Computer Association Nepal, said Nepali companies were for now “more focused on teaching existing staff to use AI tools than on cutting headcount”. That protects people who already have a job, and does little for the next batch of graduates.
The work is staying in South Asia, and changing
A growing fear among countries selling IT services is that foreign clients may increasingly use AI to handle the work themselves. The World Bank’s October 2026 report on South Asia says outsourcing firms that specialise in lower-value tasks, “such as customer support and routine software development”, are particularly exposed. It also reports that Indian suppliers of AI-exposed services cut hiring sharply after ChatGPT arrived, and that the cuts appear to reflect foreign buyers automating tasks they used to outsource.
The same report also offers a promising picture. Business links between US customers and South Asian suppliers have grown fastest in the sectors most exposed to AI. The World Bank concludes that clients taking work back is not the main trend so far. Instead, demand is shifting to higher-value work that sits next to the tasks AI automates. One example it gives is a growing market for “expert human review of AI-generated software and text”.
Where Nepal lands in that shift depends largely on its freelancers. A study by Institute for Integrated Development Studies (IIDS) found that freelancers accounted for about 61% of IT export earnings in 2022, with 14,728 freelancers working in software development and technology. A freelancer has no employer to retrain from, and a client who once paid a freelancer in Kathmandu to build a basic website can now get a working first draft from an AI tool.
We may not see the effect in the data for years. When official figures say NRs 22 billion and the industry says NRs 145 billion, a fall in small freelance contracts can pass unnoticed. NAS-IT has said its new survey will be ready by the end of this year. It must ask how much entry-level work has been lost to AI tools.
The skills likely to hold their value
Nepal’s IT graduates faced skill gaps before AI arrived. In the IIDS survey, 63% of IT companies rated the skills of their entry-level employees as below average, and 70.4% said a fresh hire needed six months of training to become competent. According to Abhay Poudel, treasurer of NAS-IT, his company hires ten to twelve graduates in a year, with each requiring four to six months of basic training.
AI raises that bar. In Stack Overflow's 2025 survey, 84% of developers said they use or plan to use AI tools, but 46% distrust the accuracy of the output and only 33% trust it. Two in three said their biggest frustration is AI answers that are “almost right, but not quite”, and 45% said debugging AI-generated code takes more time.
Someone has to catch the “almost right”. A junior who can read code they did not write, test, find the security hole and explain to a client why the generated version will fail is worth more than one who can only produce code quickly. Giri said roles in product management, UI/UX design, quality assurance, cloud engineering and network administration have held steadier. Ankur Sharma of Leapfrog Technology shared that companies now need “people with a software engineering foundation who can use AI to make their work better”. The World Bank found that jobs asking for AI skills pay nearly 30% more than other professional roles.
What employers, colleges and graduates can do
The easy response for an employer is to stop hiring juniors and give a few seniors AI tools. That works for a year or two, but today's seniors learnt their trade on the routine work that is now shrinking, and a company that hires no juniors will have no seniors in five years. The Microsoft and Accenture trials suggest another route. Juniors gained the most from AI help, so the same tools can bring new staff up to speed faster.
Colleges can change what they mark. Working code proves little now that a tool can produce it. Students can be asked to review code they did not write, find its bugs, write tests for it and defend their choices in person. This is slower to grade, and much closer to what a new hire will do.
For graduates, the first rung of the ladder, writing simple code to a clear brief, is getting narrower. The way up runs through judgement: knowing whether code is correct, safe and right for the problem. Good developers have always needed that. New ones now need it from their first day.
Read More Stories
Meat inspection law comes into force in Kathmandu Valley, requiring official stamps on meat
The Animal Slaughterhouse and Meat Inspection Act, 2055, has come into effect in...
NEPSE falls 12.79 points as turnover climbs to Rs 4.73 billion
The stock market declined by double digits on Tuesday as the Nepal Stock...
Government honours 19 top taxpayers across different business sectors
The government has honoured 19 companies and individuals for paying the highest amounts...