Sedifly CEO Warns of AI Infrastructure Bubble

Joash Lee, CEO of Sedifly, describes the artificial intelligence market as experiencing a simultaneous infrastructure build-out super-cycle and an early-stage industrial shift in the application layer. While infrastructure providers like Nvidia report massive growth, including first-quarter fiscal 2027 revenue of $81.6 billion, Lee warns that a gap between high capital expenditure and realized enterprise return on investment creates bubble-like characteristics.

Lee supports this warning with data from the MIT NANDA initiative, which found that 95% of generative AI pilots fail to deliver measurable profit-and-loss impact. This aligns with a 2025 McKinsey & Company survey indicating only 37% of organizations attribute any EBIT impact to AI.

Conversely, Lee suggests that the application layer provides more durable opportunities. He advises founders to prioritize depth over breadth by embedding AI into regulated, high-stakes vertical workflows in sectors such as healthcare, finance, law, and education. He argues that these specific industry problems are where general-purpose models struggle most with liability and proprietary data.