OpenAI Forecasts Massive Revenue, Burn

- OpenAI's annualized revenue reached $13.1 billion in 2025, a significant increase from $3.7 billion in 2024, driven by ChatGPT subscriptions and a growing enterprise customer base. - The company is reportedly finalizing a funding round of over $100 billion, which could push its valuation to approximately $850 billion, with major backing from tech giants like Microsoft, Nvidia, and Amazon. - A primary driver of OpenAI's expenses are "inference costs"—the computing power required to run AI models for users—which quadrupled in 2025 and caused adjusted gross margins to fall from 40% to 33%. - The cost of training new, more powerful AI models has been escalating exponentially; while GPT-3 cost an estimated $4 million to train in 2020, GPT-4's training cost is estimated to be over $100 million. - To secure the immense computing power it needs, OpenAI has struck multi-year deals with infrastructure providers, including a commitment to pay Oracle $60 billion annually for five years for cloud infrastructure. - OpenAI has revised its long-term spending plans, reducing its projected compute spending by 2030 from $1.4 trillion to a more conservative $600 billion. - The company's spending is projected to escalate significantly in the coming years, with expectations to reach $17 billion in 2026 and $45 billion in 2028, primarily for expanding compute infrastructure and developing proprietary AI chips. - Despite its rapid revenue growth, OpenAI does not expect to be cash-flow positive until 2030, when it anticipates generating around $40 billion in cash.