The rapid development of AI tools has shifted the bottleneck in the IT industry from code production to the validation of viable business ideas. While AI allows for the creation of MVPs in weeks rather than quarters, many of these products fail to find market demand, highlighting a lack of profitable concepts rather than a lack of engineering capacity.
Data from Sensor Tower indicates that 600,000 new applications were released in the App Store in 2025, a 30% year-over-year increase. In the first half of 2026, another 560,000 apps were launched. Despite this surge in supply, download growth remained stagnant at 3% in 2025 and 2% in the first half of 2026, suggesting that many new projects lack real utility.
Garry Tan, CEO of Y Combinator, noted in March 2026 that startups in their batches frequently produce 10,000 to 20,000 lines of code per day. Consequently, the accelerator is shifting its focus away from traditional credentials toward product intuition and agency. Y Combinator now prioritizes niche professional software over generic AI-based projects.
The industry is currently moving toward 'Small Software'—tools built by individuals or small groups that were previously too costly to develop. This trend challenges the traditional SaaS model, as clients can now generate custom solutions on demand rather than purchasing off-the-shelf software. Companies are cutting payroll expenses to avoid funding projects that fail to generate revenue.
In a separate development, Bloomberg employee Carter Leffen used the GPT-6 Astra model to decrypt a German military radio message from 1941. The message, which had resisted decryption attempts since 2005, was cracked by the AI after approximately ten hours of processing time during a study conducted on September 14 and 15.
