A recent report titled "State of AI: Global Survey 2026," published by the consulting firm McKinsey on August 25, 2026, reveals that nearly a third of companies have stopped purchasing at least one software or feature. This decision is largely due to the growing use of AI-powered coding tools that allow companies to develop these functionalities internally. The survey, conducted from May 4 to June 8, included 1,719 respondents from 97 countries, and highlights a shift in how businesses are managing their technology needs. The report shows that companies with the largest engineering teams and the most advanced AI infrastructure are most likely to discontinue software subscriptions. These cancellations are visible in budgets, but the replacement—internal development—remains largely invisible for several quarters. This shift could pose hidden risks, particularly in terms of security. As companies move away from third-party software, they may face challenges in maintaining the same level of security and compliance as before. Microsoft has been a vocal proponent of this transition, stating in August 2025 that traditional enterprise software, such as SaaS applications, are becoming obsolete. The company predicts that AI agents will replace these systems by 2030. According to Microsoft, these AI agents will be able to adapt dynamically to user needs, reducing the need for traditional enterprise applications. Charles Lamanna, a Microsoft executive, has likened these traditional applications to the mainframes of the past, suggesting they will become relics by the 2030s. Despite the growing adoption of AI agents, the transition is not without challenges. McKinsey’s survey notes that while 37 percent of respondents report that AI has positively impacted their EBIT (Earnings Before Interest and Taxes), about 80 percent report individual productivity gains. However, about 20 percent of companies limit AI use due to operational costs, and approximately 10 percent have already reached a cost ceiling for their AI tools. The cost of deploying AI agents can be high, with estimates ranging from $20,000 to $300,000 per workflow, depending on the complexity of the system. This suggests that while the shift to AI is happening, it is not yet universal and is unevenly distributed across different company sizes and sectors.