Марія БровінськаAI Eng
24 July 2026, 08:23
2026-07-24
"Selling the hours of a developer who has become three times more efficient thanks to AI is pure business masochism." InSoft.Partners CEO on how artificial intelligence is destroying classic IT outsourcing
Head of the InSoft.Partners holding company, which invests in the service business, Vitaliy Gorovyi: the classic IT outsourcing business model is not experiencing temporary stagnation due to geopolitics or mobilization, but a much deeper structural breakdown. Generative artificial intelligence destroys the very logic of labor arbitrage, on which the industry has been based for decades.
Head of the InSoft.Partners holding company, which invests in the service business, Vitaliy Gorovyi: the classic IT outsourcing business model is not experiencing temporary stagnation due to geopolitics or mobilization, but a much deeper structural breakdown. Generative artificial intelligence destroys the very logic of labor arbitrage, on which the industry has been based for decades.
The author in a column for Forbes Ukraine describes the global outsourcing market as a field divided between four regional clusters: India and the Philippines hold the mass segment thanks to an army of coders and rates of $15–50/hour; Latin America specializes in nearshoring for the USA due to the coincidence of time zones and charges $35–80/hour; Central and Eastern Europe is a premium R&D zone with strict EU compliance and rates of $40–75/hour; Vietnam, with over 530,000 developers, acts as an aggressive budget player with minimal staff turnover. According to Horovoy, it is precisely into this established field that the AI factor has «taken off,» devaluing the advantages of both a low and a high check.
The businessman identifies three simultaneous AI blows to the classic outsourcing model.
Economic paradox. The service business has traditionally earned money using the Time & Material model — payment for developer hours. If AI assistants allow an engineer to make a feature three times faster, the company using the old model sells the client three times fewer hours. According to the author’s observation, large customers are already directly demanding the so-called AI discount when signing contracts — on average, 15–25% of standard rates.
Transformation of the personnel pyramid. Juniors have suffered the most: autonomous AI agents write template code and do basic testing themselves, so the threshold for entering the profession has risen sharply. Weak middlemen, who worked as «live converters» of technical requirements into code, are being washed out by the market, while talented middlemen armed with AI now produce senior-level results for a lower price. Gorovy warns of the long-term effect: companies that massively refuse to hire beginners are themselves blocking the flow of personnel, and in three to five years the industry may face a personnel vacuum when the current seniors go into management or retire, and there will be no one to replace them.
New AI compliance. Enterprise customers — banks, healthcare, large retailers — no longer accept code written using public AI tools without guarantees. Contractors must prove that developers did not «feed» proprietary code to public models, and that the generated AI code does not violate other people’s licenses — otherwise the company is automatically eliminated from international tenders.
How to survive: the new matrix
The author sees two areas of salvation for service companies. The first is the transition from Time & Material to Outcome-Based or Value-Based Pricing, when the client pays not for hours, but for a specific result and the speed of the feature’s release to the market, and the company keeps all the AI efficiency as a margin. The second is the rejection of the «universal gallery» model, which takes on any development, in favor of either narrow hyper-specialization in complex areas where AI will not replace humans for a long time (cybersecurity, complex cloud infrastructure, embedded systems, banking architecture), or Asset-Backed Services — when the company comes to the client with ready-made product blocks and frameworks, and AI only quickly customizes them for a specific task.
