The profitability of AI models, especially OpenAI's GPT-5, is under scrutiny. While Sam Altman and Dario Amodei claim that AI models can cover their own R&D costs, the reality is more complex. OpenAI reportedly generated $6.1 billion in revenue during GPT-5’s four-month lifespan. However, the costs of running the model, including inference compute, staff compensation, sales and marketing, and administrative expenses, totaled around $6.8 billion. This suggests an operating loss of $0.7 billion, indicating that while gross margins are decent at approximately 48%, the overall operating margins are negative. The article dives deeper into the lifecycle profitability of AI models. OpenAI's estimated R&D expenditure for GPT-5 is around $5 billion, which is more than the gross profits generated by the model. If AI models like GPT-5 are replaced quickly by competitors, they may not generate enough revenue over their short lifetimes to recoup their development costs. OpenAI's partnership with Microsoft complicates matters further, as 20% of its revenue goes to Microsoft, which adds pressure to profitability. The analysis reveals that while AI models might show promise in gross profitability, their overall financial health is questionable when considering all associated costs.