OpenAI’s rising costs challenge ambitions despite margin surge
OpenAI has sharply improved its compute margins, doubling them from around 35 percent in early 2024 to nearly 70 percent by late 2025. The improvement, driven by optimized model performance, reduced reliance on rented computing resources, and the introduction of premium subscription tiers, underscores the company’s push toward profitability amid rising competition and investor scrutiny.
Despite these operational gains, OpenAI continues to face significant financial pressures. The company generated around 4.3 billion dollars in revenue during the first half of 2025, surpassing its entire 2024 performance by 16 percent. Yet, high infrastructure spending and massive capital requirements have left it deeply unprofitable, with an estimated cash burn approaching 9 billion dollars for the year.
Efficiency gains amid industry competition
OpenAI’s improved compute efficiency now exceeds that of its closest competitor, Anthropic, whose margins are forecast to exceed 50 percent by the end of 2025. However, in absolute spending terms, Anthropic remains more cost-efficient. Forecasts suggest its compute expenditure will reach about 6 billion dollars in 2025, compared with OpenAI’s 15 billion, with the gap expected to widen over the coming years.
The company launched a new premium ChatGPT Pro plan at 200 dollars per month in December 2024, further boosting profitability. However, analysts note that the recent surge in resource optimization followed an internal “code red” triggered by the release of highly efficient models by rivals such as DeepSeek.
Valuation pressures and infrastructure spending
In October 2025, OpenAI completed a secondary share sale valuing the company at 500 billion dollars, making it the most valuable privately held company in the world. Investors including SoftBank, Thrive Capital, and MGX from Abu Dhabi purchased 6.6 billion dollars in existing stock from employees and early backers.
Despite record valuations, OpenAI faces a long-term financial challenge. It is projected to face a funding gap of over 200 billion dollars by 2030 due to its heavy investment commitments. The company has allocated more than 1 trillion dollars in hardware and cloud infrastructure spending through 2035, involving major partners such as Oracle, Microsoft, Broadcom, and Nvidia. Executives aim to reach cash flow positivity by 2029.
The improvements in efficiency mark a milestone, but OpenAI’s path to sustainable profitability remains fraught with the immense costs required to lead in the artificial intelligence arms race.
