Two prominent players in the AI industry, OpenAI and Anthropic, are encountering substantial hurdles on their journey toward profitability, as both have fallen short of their predetermined gross profit margin objectives. The main culprit is the unexpected surge in inference costs, a trend that becomes even more pronounced amidst a steadily growing user base. This has led to external skepticism regarding their ability to attain a gross profit margin exceeding 60% by the end of this decade. Last year, OpenAI saw its gross profit margin decline to 33%, falling below its projections. Meanwhile, Anthropic projects a gross profit margin of 40% for 2025, which also falls short of the desired target. Inference costs have emerged as the primary factor exerting pressure on gross profit margins. OpenAI's inference costs soared approximately four times compared to the previous year, while Anthropic anticipates a more than threefold increase in 2025. Furthermore, OpenAI grapples with gross profit margin pressures stemming from its sizable non-paying user base and product composition. Nevertheless, the efficiency of its paying users has improved, and the company plans to further refine its revenue structure. In the long run, OpenAI anticipates that the share of inference costs for paying users will climb to roughly 94% by 2030, with a gross profit margin target hovering around 67%. However, achieving this goal amid escalating costs remains a fundamental challenge. Given the market risks, investors are advised to proceed with caution.