OpenAI Posts $20.92 Billion 2025 Operating Loss on $13.07 Billion Revenue
Updated
Updated · Fortune · Jun 16
OpenAI Posts $20.92 Billion 2025 Operating Loss on $13.07 Billion Revenue
3 articles · Updated · Fortune · Jun 16
Summary
$20.92 billion in operating losses swamped OpenAI’s $13.07 billion of 2025 revenue in leaked financial statements, even as sales more than tripled from $3.7 billion a year earlier.
$34 billion in total costs drove the deficit, led by $19.18 billion in R&D, $7.5 billion in cost of revenue, and $5.73 billion in sales and marketing.
The leak suggests OpenAI is still buying growth ahead of an expected IPO later this year, though its efficiency improved: expenses fell to $1.60 per $1 of revenue from $2.37 in 2024.
Profitability would likely require either higher prices or new revenue streams, while cutting R&D remains difficult because model performance is central to retaining customers.
With losses soaring towards $85 billion, are OpenAI's new mega-deals a true lifeline or just delaying an inevitable financial collapse?
Founded to prioritize safety, has OpenAI's relentless pursuit of capital now sacrificed its mission for the sake of market dominance?
As AI's energy demand rivals entire nations, is the multi-trillion dollar infrastructure plan for AGI environmentally or economically sustainable?
OpenAI’s $4 Billion Revenue, $6.7 Billion R&D Spend, and the High-Stakes Gamble for AI Supremacy in 2025
Overview
In 2025, OpenAI reached an estimated $4 billion in annual revenue, marking a major milestone as it pursues a bold goal of $100 billion by 2029. However, this growth came with massive R&D spending—$6.7 billion in just the first half of the year—and resulted in substantial financial losses. Despite holding $17.5 billion in cash and securities, OpenAI expects to face significant annual losses until at least 2030. The company’s aggressive investments reflect the high costs typical across the AI industry, raising important questions about the long-term sustainability and transparency of its business model.