Algorithmic fairness and data bias are heavily debated in the medical AI community. However, these discussions assume an existing diagnostic infrastructure. A more fundamental challenge is that 47% of the world's population does not have access to basic diagnostics. For these groups, improving diagnosis with AI is not the main concern. The existence of the diagnostic capacity itself is key. The current trajectory of AI innovation threatens to widen this diagnostic gap. The article confronts this global inequality in healthcare.