华尔街见闻·2025-10-09 11:14Core Viewpoint - OpenAI has signed nearly $1 trillion in computing power procurement agreements, significantly exceeding its revenue and financing capabilities, raising questions about how the company can support such a massive gamble [3][5][7]. Group 1: Financial Agreements - OpenAI's agreements include $500 billion with Nvidia, $300 billion with AMD, and $300 billion with Oracle, along with over $22 billion with CoreWeave, aiming to provide over 20 GW of computing power over the next decade [7]. - The company is utilizing a "circular financing" model and an innovative "equity-for-purchase" approach to secure funding from suppliers, effectively rewriting capital rules in the AI era [5][8]. Group 2: Financial Models - The AMD model involves a groundbreaking agreement where OpenAI plans to purchase up to $90 billion worth of AMD GPUs while receiving warrants to buy up to 160 million shares at $0.01 each, potentially allowing OpenAI to acquire significant equity at minimal cost [10]. - The Nvidia model involves a direct investment of up to $100 billion, which provides OpenAI with cash to purchase Nvidia chips, creating a "circular revenue" flow where funds are reinvested back into purchasing [11][12]. Group 3: Financial Analysis - Goldman Sachs estimates that OpenAI's operational infrastructure costs will reach approximately $35 billion by 2026, with self-generated revenue contributing about 48% [15]. - However, when considering future capital commitments, total funding needs could soar to around $114 billion, with external equity and debt financing needs rising to 75% of the total funding structure [17][19]. Group 4: Market Impact and Risks - The partnerships have significantly boosted the market valuations of companies like Oracle and AMD, with market caps increasing by $244 billion and $63 billion, respectively, following the announcements [22]. - Despite the innovative financial structures, OpenAI's reliance on external capital raises concerns about sustainability, especially if user growth or willingness to pay slows down [25][21].