Blab Tech
OpenAI Reports $38.5 B Net Loss in 2025, Highlights Rising Compute Costs
OpenAI disclosed a $38.53 billion net loss for the 2025 fiscal year, according to audited financial documents reviewed by Ed Zitron’s Where’s Your Ed At and independently shown to the Financial Times. The loss follows a revenue increase to $13.07 billion, up from $3.7 billion in 2024, and marks a sharp rise in operating costs.
The company’s 2025 revenue of $13.07 billion represents a 16 % year‑over‑year increase, matching the growth reported for the first half of the year when OpenAI earned roughly $4.3 billion. However, total costs and expenses climbed to $34 billion, a 37 % jump from the $20.92 billion recorded in 2024. The resulting net loss attributable to the company was $38.53 billion, nearly eight times the $5.09 billion loss reported for 2024.
Breakdown of the 2025 cost categories shows that the line item for cost of revenue—expenses related to serving model outputs—rose from $2.65 billion in 2024 to $7.5 billion. Research and development costs more than doubled, increasing from $7.81 billion to $19.18 billion. Sales and marketing spend also accelerated, growing from $1.11 billion to $5.73 billion. These figures illustrate how the capital intensity of frontier inference has expanded as newer, larger models require more compute to train and to serve.
Inference spending on Microsoft Azure was a key driver of the 2025 cost surge. OpenAI paid $5.02 billion for inference services on Azure in the first half of 2025 alone. For the period covering calendar year 2024 through the third quarter of 2025, the total inference spend on Azure reached $12.43 billion, according to the documents. The figures cover only inference, not the training of models, meaning the full compute bill is larger.
OpenAI’s rapid top‑line growth has been matched by an even faster expansion of its cost base. Each new model iteration demands more GPU power, power consumption, and data‑center capacity, all of which do not become cheaper as demand rises. The company’s reliance on Azure for compute resources underscores the strategic partnership with Microsoft, which has invested over $13 billion in OpenAI and provides the cloud infrastructure.
The financial condition of OpenAI has drawn concern from analysts. Zitron noted that the $38.53 billion loss is “astronomical” and that losses appear to be mounting year‑over‑year at a dramatic rate. OpenAI declined to comment to the Financial Times.
In related reporting, OpenAI’s chief financial officer, Sarah Friar, said in a blog post that the company’s annualized revenue had surpassed $20 billion in 2025, up from $6 billion in 2024. Friar also indicated that profitability remains far away, a view echoed by the 2025 financials that show a $13.5 billion loss in the first half of the year.
OpenAI’s 2025 financials come amid a broader industry trend of escalating AI compute costs. The company’s compute capacity reached 1.9 GW in 2025, up from 0.6 GW in 2024, and it announced a $40 billion funding round at a $300 billion post‑money valuation in March 2025. The company’s restructuring in 2025, which converted its for‑profit subsidiary into a public‑benefit corporation, and a $6.6 billion share sale that valued the company at $500 billion, also highlight the scale of capital deployed.
At present, OpenAI has not outlined a path to profitability or detailed cost‑reduction measures. The company’s financial statements show that while revenue is growing, the cost of delivering and training large language models continues to outpace that growth. Investors and industry observers will be watching for any signals that OpenAI can curb its compute expenditures or shift its business model toward more sustainable margins.
The company’s 2025 revenue of $13.07 billion represents a 16 % year‑over‑year increase, matching the growth reported for the first half of the year when OpenAI earned roughly $4.3 billion. However, total costs and expenses climbed to $34 billion, a 37 % jump from the $20.92 billion recorded in 2024. The resulting net loss attributable to the company was $38.53 billion, nearly eight times the $5.09 billion loss reported for 2024.
Breakdown of the 2025 cost categories shows that the line item for cost of revenue—expenses related to serving model outputs—rose from $2.65 billion in 2024 to $7.5 billion. Research and development costs more than doubled, increasing from $7.81 billion to $19.18 billion. Sales and marketing spend also accelerated, growing from $1.11 billion to $5.73 billion. These figures illustrate how the capital intensity of frontier inference has expanded as newer, larger models require more compute to train and to serve.
Inference spending on Microsoft Azure was a key driver of the 2025 cost surge. OpenAI paid $5.02 billion for inference services on Azure in the first half of 2025 alone. For the period covering calendar year 2024 through the third quarter of 2025, the total inference spend on Azure reached $12.43 billion, according to the documents. The figures cover only inference, not the training of models, meaning the full compute bill is larger.
OpenAI’s rapid top‑line growth has been matched by an even faster expansion of its cost base. Each new model iteration demands more GPU power, power consumption, and data‑center capacity, all of which do not become cheaper as demand rises. The company’s reliance on Azure for compute resources underscores the strategic partnership with Microsoft, which has invested over $13 billion in OpenAI and provides the cloud infrastructure.
The financial condition of OpenAI has drawn concern from analysts. Zitron noted that the $38.53 billion loss is “astronomical” and that losses appear to be mounting year‑over‑year at a dramatic rate. OpenAI declined to comment to the Financial Times.
In related reporting, OpenAI’s chief financial officer, Sarah Friar, said in a blog post that the company’s annualized revenue had surpassed $20 billion in 2025, up from $6 billion in 2024. Friar also indicated that profitability remains far away, a view echoed by the 2025 financials that show a $13.5 billion loss in the first half of the year.
OpenAI’s 2025 financials come amid a broader industry trend of escalating AI compute costs. The company’s compute capacity reached 1.9 GW in 2025, up from 0.6 GW in 2024, and it announced a $40 billion funding round at a $300 billion post‑money valuation in March 2025. The company’s restructuring in 2025, which converted its for‑profit subsidiary into a public‑benefit corporation, and a $6.6 billion share sale that valued the company at $500 billion, also highlight the scale of capital deployed.
At present, OpenAI has not outlined a path to profitability or detailed cost‑reduction measures. The company’s financial statements show that while revenue is growing, the cost of delivering and training large language models continues to outpace that growth. Investors and industry observers will be watching for any signals that OpenAI can curb its compute expenditures or shift its business model toward more sustainable margins.
