Technological dependency has become a growing strategic risk for small and medium-sized enterprises (SMEs). Many SMEs are not fully aware of how much they rely on specific technologies. For years, their focus has been on ensuring their tools work correctly, rather than considering how easily they could switch or stop using them. A business software developed fifteen years ago, which has been adapted over time and is now mastered by a few key employees, can continue to serve its purpose while creating a significant dependency. As long as the publisher remains stable and commercial conditions change little, this vulnerability often goes unnoticed. It becomes more apparent when a supplier alters its conditions, business model, or licensing policy. When a tool contains years of business rules, data, and processes within proprietary technology, the ability to change solutions becomes a strategic concern. WinDev, a business software, illustrates this problem well. The recent situation around PC Soft, its publisher, provides a particularly clear example. After PC Soft was acquired at the end of 2024 by a company linked to Volaris Group, a subsidiary of Constellation Software, the licensing model for WinDev changed significantly. Perpetual licenses disappeared, the annual developer cost increased by 43 percent, and a fee for deployed applications was introduced. This change triggered several investigations, a petition signed by more than 1,600 people, and two written questions to the National Assembly in June 2026. The issue raised by this case goes beyond PC Soft. It highlights how a company might only realize late that it is heavily dependent on a technology it has used for a decade or more. General management typically knows the annual cost of their software but has little idea of how much time and money it would take to replace it. This "exit cost" is a key measure of real dependency. Artificial intelligence is now beginning to change this dynamic. Until recently, exiting a legacy system could take months or even years—just to understand the existing code, recover business rules, map dependencies, and secure a rewrite. This difficulty explained why many modernization projects were delayed. Artificial intelligence is now helping to speed up this process. It can analyze code faster, identify dependencies, generate tests, and even assist in rewriting parts of the system. It does not replace human expertise or architectural planning but significantly reduces the time needed to understand an old system. As a result, migrations that were once thought to take years can now be completed in a few months. For SMEs, the challenge goes beyond a single publisher or technology. Digital sovereignty does not mean banning proprietary solutions but maintaining the ability to choose. A company should be able to answer this question: if our supplier suddenly changed its conditions tomorrow, how long would it really take us to leave?