For nearly two decades, software development outsourcing has been a fast and cost-effective way to access global talent and scale. By 2024, the IT outsourcing market exceeded $512 billion, with companies saving up to 70% on labor and gaining flexibility by outsourcing teams to India, Eastern Europe, and Latin America (and, of course, Ukraine).
For nearly two decades, software development outsourcing has been a fast and cost-effective way to access global talent and scale. By 2024, the IT outsourcing market exceeded $512 billion, with companies saving up to 70% on labor and gaining flexibility by outsourcing teams to India, Eastern Europe, and Latin America (and, of course, Ukraine).
Outsourcers provided (and continue to provide) jobs for tens of thousands of Ukrainians. They still occupy the largest number of places in the top 10 IT companies in Ukraine. But the peak of the number of personnel in most of them fell on the end of 2021-beginning of 2022. Yes, of course, the war made very noticeable adjustments to the number and quality of personnel. But there was and remains another very important lever.
A recent report from IT research firm Jellyfish claims that nearly two-thirds (64%) of companies generate the majority of their code using artificial intelligence. What’s more, if current trends continue, this share could reach 90% within a year.
«AI-powered coding tools are now the default option for engineering teams, and the productivity gains are real,» writes Nicholas Arcolano, head of research at Jellyfish.
Analysts in the largest outsourcing markets are beginning to fear that rapidly developing artificial intelligence tools could literally turn the IT sector upside down. In India, it is estimated at $283 billion and depends largely on cheap human resources… which can be automated.
«The market fears that AI tools could replace IT services that are currently outsourced. What the real impact will be remains to be seen,» said V. K. Vijayakumar, chief investment strategist at Geojit Investments (an Indian investment firm).
According to the NBU, export revenue from computer services in Ukraine increased by 3,3% in 2025 and reached $6.66 billion. This is ten times less than the Indian market. But this is a fairly significant export component of the economy during the war. If it starts to melt, it will be noticeable for both the local labor market and the country’s budget.
Foreign customer companies are increasingly starting to think about expanding the power of their own IT and product teams thanks to the capabilities of AI tools. Jellyfish researchers say that development efficiency can grow (as respondents expect) by 25% per year. In other words, they can do more work with fewer resources. This adds risks to this part of the business that is outsourced. Yes, you can save more on it, but it also creates typical problems.
Here’s how a service technician from a Western cybersecurity company recalls his journey with outsourcing companies:
«Like many young companies, we relied on external vendors to scale quickly. However, the drawbacks began to show. Projects managed by external teams often suffered from loss of context, inconsistent standards, and delays in learning cycles. What looked efficient on paper turned out to be expensive in practice—cheap modules delivered quickly, followed by months of debugging and integration rework.»
But the market is changing rapidly. And tools like Copilot, Cursor, Augment, Gemini Code Assist, Amazon Q, Lovable, Codex, Kilo are markers of a changing era. Moreover, they are starting to be used not only by engineering teams, but also by «humanitarians» who have never written code before. So, now we are witnessing a boom in vibecoding, which is creating additional turmoil in the market.
Tasks that were previously assigned to junior developers or outsourced teams—debugging, testing, documentation, boilerplate code—can now be done faster and sometimes even more efficiently thanks to AI.
Agent coding workflows (autonomous AI agents that can plan, write, and test code without constant human prompting) and AI copilots (assistant coding tools that suggest, generate, and optimize code in real time) operate continuously, learning from repositories and internal task sets, knowledge bases.
They don’t wait for data to be transferred, they don’t lose context, and they don’t bill by the hour. They’re not people who sit across the ocean and sometimes spend months delving into the context of projects.
If the prediction of 90% of AI-generated code is realized, software engineering could move from artisanal logic to algorithmically driven manufacturing, where autonomous agents will play an increasingly important role.
Another challenge: there may be a lot of low-quality code on the market.
«Younger/less skilled engineers/people with no software development experience can generate 10x more buggy/underperforming code thanks to AI. The amount of code generated by AI can create a maintenance/support capacity issue until we get to a point where the entire ecosystem is well enough integrated that AI can analyze and understand production issues and match them to potentially buggy code,» says one of the British engineers in the Jellyfish study.
But, most likely, this can be fixed in the future without manual intervention. The key conclusion: such market changes will lead to cheaper code. And if it becomes cheaper quickly, then the market will change radically.
