Chainfeeds Summary:
OpenAI can never go back to what it was before, but perhaps it never intended to.
Article source:
https://mp.weixin.qq.com/s/STX1FTMPn_p5B37-ZJ3EhA
Article Author:
Appso
Opinion:
Appso: The structure of this funding round is essentially more like a binding of interests around the computing power supply chain than a simple bet on the future of an AI company. Amazon, NVIDIA, and SoftBank have each invested tens of billions of dollars, ostensibly to support OpenAI, but the ultimate destination of the funds is highly predictable: purchasing GPUs, leasing cloud services, and expanding computing infrastructure. In other words, these giants are not only investors but also the primary revenue recipients in the future, effectively securing the world's largest computing power customers in advance. This investment-and-return model makes the funding round seem like a low-risk business. Meanwhile, the capital structure is also changing, from ARK Invest to traditional institutions, and then to high-net-worth individuals entering through banking channels. OpenAI is gradually opening up broader funding channels, preparing for a potential IPO. From a data perspective, the company's growth remains strong: nearly 900 million weekly active users, over 50 million paying users, and annual revenue of $13.1 billion, with a peak monthly revenue of $2 billion. The key issue, however, is that it has yet to achieve profitability, and its computing costs remain high, meaning that growth in scale has not led to a cost inflection point, foreshadowing subsequent strategic contraction. Sora's shutdown precisely reveals the core contradiction of the current AI business model: the gap between technological capabilities and economic viability. Video generation is visually impactful, but its computational consumption far exceeds that of text and code generation; each rendering represents real GPU resource usage and electricity costs. According to The Wall Street Journal, Sora consumed approximately $1 million daily, while its user base rapidly declined from 1 million at launch to less than 500,000. This means that even if a product is technically impressive, it cannot sustain long-term operation without generating a stable willingness to pay. Against this backdrop, OpenAI's strategy has become noticeably more converging: on the one hand, continuing to advance core model capabilities, such as the rapid growth of GPT-5.4 and code agents; on the other hand, concentrating resources on areas with clear commercial paths, such as enterprise services, APIs, and search. Currently, enterprise revenue accounts for over 40% and is expected to reach parity with consumer revenue. This structural shift reflects the company's move from showcasing technological possibilities to building a sustainable revenue model. Sora's exit is not an isolated case, but rather a signal: high-cost, low-monetization-efficiency products will be gradually phased out. From a longer-term perspective, OpenAI's transformation can be understood as moving from creating miracles to becoming infrastructure. In its early days, the company emphasized using general artificial intelligence to benefit all of humanity, but as commercialization progresses, its positioning is shifting towards intelligent utilities—becoming an indispensable provider of underlying capabilities for enterprises and developers. This is particularly evident in its super-application strategy: integrating ChatGPT, code generation, search, and browsing capabilities into a unified entry point, lowering the barrier to entry, and driving enterprise procurement through consumer usage habits. In contrast, individual user subscription behavior fluctuates significantly, while enterprise clients, once their core business is built on the model, face extremely high migration costs and stronger stickiness. This is precisely the revenue structure that the capital market values more. At the same time, the company is also proactively shrinking exploration directions without clear return paths, reallocating resources to businesses with high certainty. The phase of constantly releasing amazing demos in recent years may be coming to an end, replaced by an operational logic closer to traditional industries: controlling costs, optimizing resource allocation, and strengthening sales and distribution capabilities. OpenAI may no longer be pursuing changing the world with every step, but rather choosing to become the underlying system that supports the world's operation.
Content source 
