MUMBAI: Software vendors may have a new rival, and it is sitting inside the enterprise. AI-powered coding agents are increasingly encouraging companies to build software themselves instead of buying products from vendors, according to a new research report by McKinsey & Company.
Nearly one-third, or 32 per cent, of respondents said their organisations had decided not to purchase at least one software product or feature because they could build the required functionality internally using agentic coding tools.
The shift was strongest in technology and healthcare, followed by professional services and energy and materials, suggesting that companies with greater digital capabilities may be among the first to rethink the traditional “buy” approach to software.
The change comes as businesses scale the use of AI agents across their operations. Among organisations with annual revenue of more than $1 billion, 40 per cent of respondents said they were scaling AI agents in at least one business function. That is up from 27 per cent a year earlier.
Smaller organisations, however, have been slower to move. Adoption among these companies remained broadly unchanged at 22 per cent, highlighting a growing gap between large enterprises and smaller businesses in the deployment of agentic AI.
Companies that McKinsey identifies as AI high performers are taking the shift even further. They were twice as likely as other organisations to report scaling software coding agents and 2.7 times more likely to be scaling other forms of agentic AI.
Their approach to software purchasing is changing too. Nearly half of AI high performers said they had chosen to build software internally instead of buying it, compared with 31 per cent of other respondents.
For technology vendors, that could become an increasingly important warning sign. As coding agents become capable of handling more complex development tasks, enterprises may find that some off-the-shelf software is no longer worth purchasing when similar functionality can be built in-house.
That does not necessarily mean businesses will stop buying software altogether. Instead, the decision is likely to become more selective, with companies weighing what they should buy, what they can build and where they need to develop internal expertise.
The economics of agentic AI, meanwhile, are becoming harder to ignore. Around one in five organisations surveyed said AI-related operating costs, including token costs, had limited their use of the technology.
That creates a new calculation for businesses. Building a product internally may eliminate a software licence or vendor contract, but it can also bring additional costs for computing, AI model usage, engineering talent and ongoing maintenance.
Despite those concerns, the appetite for investment remains strong. About 60 per cent of respondents expect their organisations to increase AI spending over the next year.
The emerging picture is therefore more nuanced than a simple shift from buying to building. Enterprises are beginning to treat AI as another part of their technology economics, balancing the cost of external software against the growing ability to create customised tools internally.
If coding agents continue to improve, the software industry’s old “buy versus build” debate could soon have a much more AI-shaped answer.
