The Rise of In-House AI Development
A striking 32% of companies surveyed by McKinsey have opted to build their own software solutions using AI agents, rather than purchasing off-the-shelf products. This trend, detailed in McKinsey's State of AI 2026 survey, indicates a fundamental shift in how organizations approach software acquisition and development. The technology sector shows an even stronger inclination, with 41% of tech companies making this choice.
This move away from traditional software procurement suggests that the capabilities of modern AI agentic tools have reached a point where they are perceived as a viable, and perhaps even superior, alternative to commercial software. Instead of evaluating vendors, negotiating licenses, and managing integrations, businesses are leveraging AI to generate custom solutions tailored to their specific needs. This could translate to significant cost savings, faster deployment times for bespoke features, and greater control over the technology stack.
Why Companies Are Building, Not Buying
Several factors likely contribute to this growing trend. Firstly, the increasing sophistication of AI coding agents means they can now handle complex development tasks. These agents can understand natural language prompts, generate code, identify and fix bugs, and even optimize performance. For companies with existing engineering talent, this allows them to redirect their developers' efforts towards building highly specific functionalities that off-the-shelf software might not offer or would require costly customization.
Think of it less like a traditional software development lifecycle and more like having a highly skilled, albeit digital, junior developer on staff who can rapidly prototype and build. The ability to define requirements in plain English and have an AI agent translate that into functional code dramatically lowers the barrier to entry for custom software development. This is particularly appealing for businesses with unique workflows or competitive advantages tied to proprietary technology.
Secondly, the economics can be compelling. While initial investment in AI tools and training might be required, the long-term cost of building and maintaining a custom solution with AI agents can be lower than perpetual licensing fees, maintenance contracts, and the hidden costs of integrating disparate commercial systems. Furthermore, by controlling the development process, companies can avoid vendor lock-in and ensure their software evolves in lockstep with their business strategy.
The survey results also hint at a potential challenge for traditional software vendors. If a substantial portion of the market is shifting towards in-house development powered by AI agents, vendors will need to adapt. This could mean offering more flexible, modular solutions, providing robust APIs for AI integration, or even developing their own AI-powered development platforms.
The Tech Sector Leads the Charge
The higher adoption rate within the tech sector (41%) is not surprising. Technology companies are typically early adopters of new tools and possess a higher concentration of engineering talent capable of leveraging AI agents effectively. For these firms, the ability to rapidly iterate on software that forms the core of their business is a critical competitive advantage. Building bespoke solutions with AI agents allows them to maintain agility and responsiveness in a fast-moving market.
This also raises an interesting question about the future of the software development job market. Will AI agents augment developers, allowing them to focus on higher-level design and strategy, or will they displace roles focused on routine coding tasks? The McKinsey survey suggests a trend towards augmentation, where AI acts as a co-pilot, accelerating the creation of custom tools.
Implications for the Future
The implications of this trend are far-reaching. For businesses, it signals a potential democratization of software development. Companies that previously lacked the resources or expertise to build custom software may now find it within reach. This could lead to a more diverse and innovative software landscape, with solutions tailored to niches previously underserved by commercial offerings.
For the broader technology industry, it underscores the rapid maturation of AI development tools. Agentic AI is moving beyond theoretical concepts and into practical, budget-impacting applications. This trend will likely accelerate as AI agents become more capable and accessible, further blurring the lines between traditional software development and AI-driven creation. The challenge for established software providers is clear: either integrate these AI capabilities into their offerings or risk being bypassed by a new wave of in-house, AI-powered development.
The McKinsey survey data, originating from their State of AI 2026 survey, provides a concrete data point that suggests a significant shift in software strategy. The question is no longer if AI can build software, but rather how much of it businesses will choose to build themselves rather than buy.