OpenAI has demonstrated remarkable ambition in its computing power investments, spending $16 billion (approximately 113.8 billion RMB) on leasing computing resources this year, equivalent to tens of millions of dollars in daily expenses. However, this is just the tip of the iceberg. According to The Information, the company plans to invest an additional $100 billion over the next five years in leasing backup servers. This budget does not include its previously announced $350 billion long-term computing power investment plan.
Behind this massive investment lies the reality of a computing power shortage. Chief Financial Officer Sarah Friar revealed that due to insufficient computing power, OpenAI has repeatedly delayed the release of new features and even proactively slowed down product performance. The $300 billion, five-year contract with Oracle further highlights its commitment to ensuring a secure computing power supply.
Based on financial data, OpenAI projects revenue of $13 billion this year (a 225% increase over last year), and has raised its 2030 revenue forecast to $200 billion. Despite this, its R&D investment will still account for nearly 50% of its total revenue, far exceeding tech giants like Microsoft (15.5%) and Meta (25%). Specifically, of the $16 billion allocated this year, $9 billion will be spent on training and $7 billion on inference. By 2030, these two expenditures will each approach $50 billion.
Faced with this massive investment, CEO Altman announced the launch of new "computationally intensive" products in the coming weeks, but these will be available only to Pro members at an additional cost. This move has sparked dissatisfaction among Plus users, and OpenAI continues to struggle to balance computing capacity expansion with commercial monetization.
