OpenAI spent $34 billion in 2025 as AI infrastructure race intensifies

OpenAI spent $34 billion last year to dominate the booming artificial intelligence market ahead of its planned initial public offering, the Financial Times reported Monday, in a disclosure that underscores the staggering capital requirements of the AI arms race.

Audited financial figures show the ChatGPT maker spent about $19 billion on research and development in 2025 and nearly $6 billion on sales and marketing, according to the report. The remaining $9 billion covered operational and infrastructure costs tied to running the world's most popular AI chatbot and training its next-generation models.

"The scale of spending reflects the reality that frontier AI is a winner-take-most game where the cost of entry is measured in tens of billions," said Rachel Kim, an analyst at Edgen. "OpenAI is burning cash to build a moat before its IPO forces it to show a path to profitability."

The $34 billion figure puts OpenAI's spending on par with the largest cloud infrastructure buildouts. Oracle, a key partner in the Stargate data center project with OpenAI, said last week it expects capital expenditures of up to $95 billion in fiscal 2027, with the massive Texas facility more than three-quarters complete within 90 days. Oracle's remaining performance obligations, a measure of future contracted revenue, reached $638 billion.

The Inference Gold Rush

OpenAI's spending validates a broader thesis driving the AI infrastructure boom: that the shift from training to inference — the process of running AI models in real-world applications — will require exponentially more computing power. Cerebras Systems, the AI chipmaker that went public in May, reported $510 million in revenue for 2025, up 76 percent year over year, and swung to a net profit of $237.8 million from a loss of $481.6 million in 2024.

Mizuho analyst Vijay Rakesh projects the fast-inference segment could grow into a $550 billion annual recurring revenue opportunity by 2030, expanding at a compound annual growth rate of 291 percent. Cerebras, whose Wafer Scale Engine is designed specifically for inference workloads, reported remaining performance obligations of $24.6 billion as of December 2025.

What the Spending Means for Investors

The disclosure raises questions about OpenAI's path to profitability ahead of its IPO. The company's $19 billion in R&D spending alone exceeds the total 2025 revenue of most public AI companies. Nvidia, whose GPUs power much of OpenAI's infrastructure, reported data center revenue of $130 billion in its latest fiscal year and trades at 35 times forward earnings. Microsoft, OpenAI's largest backer, has committed billions more to the partnership.

For investors, the key question is whether OpenAI can convert its spending into durable revenue before the IPO forces greater financial transparency. The company's ability to monetize ChatGPT and its enterprise API offerings will face heightened scrutiny as it prepares to go public, with the spending data providing the first detailed look at the economics of frontier AI.

This article is for informational purposes only and does not constitute investment advice.