OpenAI has clinched computing resource contracts valued at approximately $1 trillion this year, a sum that dwarfs its revenue and even outstrips the GDP of numerous nations.

Concurrently, this company, which is still racking up substantial losses, has seen its valuation skyrocket to $500 billion, rendering it the world's most valuable unlisted entity. Beneath these seemingly paradoxical financial metrics lies OpenAI's ambition to dominate the entire AI industry chain, along with a heated debate over whether this signifies a 'bubble or a breakthrough'.

01. Capability: How Does Loss-Making OpenAI Secure Its Funding?

Since the dawn of the year, this company at the forefront of the global AI surge has inked computing power deals totaling nearly $1 trillion at an unprecedented pace with titans like NVIDIA, AMD, and Oracle. The magnitude of these agreements far surpasses its own revenue and financing capabilities.

This scenario is enough to leave traditional financial experts agape. In the first half of 2025, OpenAI generated $4.3 billion in revenue while incurring losses of $13.5 billion. Yet, a company still incurring significant losses has managed to secure computing power procurement agreements worth $1 trillion.

So, where does the money originate? OpenAI tackles funding challenges through 'revolving financing' that tightly binds suppliers and an innovative 'equity-for-procurement' model.

The first approach is the NVIDIA model, a revolving revenue strategy. NVIDIA injects funds into OpenAI, which then utilizes this capital to place GPU orders with NVIDIA.

The second method is the AMD model, which involves equity in exchange for procurement. OpenAI plans to acquire and deploy up to 6GW of AMD Instinct series GPUs, valued at $90 billion. The crux of the agreement is that AMD issues warrants to OpenAI, enabling it to purchase up to 160 million AMD shares at a nominal price of just $0.01 per share. This model essentially transforms hardware sales into equity allocations. If AMD's stock price soars to $600 in the future, the total value of OpenAI's potential stake will reach $96 billion, roughly on par with the total value of the hardware procurement.

02. Controversy: Bubble or Breakthrough?

However, OpenAI's aggressive expansion has also ignited widespread controversy. Elon Musk has directly asserted that OpenAI's valuation is excessively inflated and, along with his company xAI, has filed a lawsuit against OpenAI, accusing it of stealing trade secrets.

Deeper concerns arise from the practical application bottlenecks in the AI realm. Research from the Massachusetts Institute of Technology indicates that 95% of corporate AI investments have failed to yield quantifiable benefits. A substantial amount of capital is flowing into inefficient 'AI tourism' projects.

Moreover, technological bottlenecks persist. The 'hallucination rate' (the likelihood of generating incorrect information) of AI models remains a hurdle, limiting their application in high-risk scenarios.

Historically, never has such a colossal amount of capital been poured into a technology in such a brief period. Despite AI's immense potential, its viability as a stable and profitable business model remains largely unproven.

It's worth noting that a September report by Bain & Company stated that to fund the computing power needed to meet expected demand by 2030, AI companies will collectively need to generate $2 trillion in annual revenue. However, Bain & Company predicts an $800 billion revenue shortfall compared to this level.

Against this backdrop, concerns about an AI investment bubble are intensifying. Even Altman has acknowledged the risk of a bubble multiple times in recent months, while still maintaining optimism about the technological prospects. 'Overall, are investors currently overly enthusiastic about AI? In my view, yes,' he remarked in August. 'Is AI the most significant development in quite some time? My answer is also yes.'

Amazon founder Jeff Bezos, on the other hand, believes that the current investment frenzy in the AI field is a 'beneficial bubble.' He argues that even if this bubble ultimately bursts like the internet bubble in 2000, causing stock prices to plummet, the long-term societal benefits will be immense. Bezos cited two examples: first, the massive investment in fiber-optic cables during the internet bubble, which remained as infrastructure after the bubble burst and laid the groundwork for subsequent internet development; second, during the biotechnology boom in the 1990s, although many companies failed, numerous 'life-saving drugs' were ultimately developed.

03. Ambition: From Model Provider to Full-Stack Ecosystem Architect

Behind the decision to sign these orders despite facing bubble-related doubts lies OpenAI's even grander strategic transformation—from a mere model provider to a full-stack AI ecosystem architect.

The OpenAI DevDay developer conference on October 6 clearly showcased this ambition. At the application layer, OpenAI introduced the Apps SDK, akin to an 'AI version of the iOS framework,' enabling developers to build real, interactive applications within ChatGPT. Live demonstrations illustrated how users could directly invoke Canva to generate posters and seamlessly switch to the Zillow app to search for properties within ChatGPT.

At the development tool layer, AgentKit empowers developers to design complex workflows through a visual canvas without writing code from scratch. At the model layer, GPT-5 Pro was officially made available to all developers via API, and Sora 2 also opened up video generation capabilities to developers for the first time.

Therefore, from custom chips at the bottom layer (in collaboration with Broadcom) to physical data centers and energy facilities in the middle layer (the 'Stargate' plan), and then to AI models and applications at the top layer, OpenAI is constructing a completely self-sufficient and highly optimized ecosystem.

In this context, OpenAI was previously perceived as a 'model company.' Its core capability was training the technology behind ChatGPT and Sora. However, Altman stated bluntly in this interview: To create truly useful AI, relying solely on models is insufficient. We must build our own infrastructure and control how users interact with AI.

Ultimately, this trillion-dollar gamble aims for far more than just developing a better chatbot. It represents an epic investment aimed at constructing the foundational platform for the next technological era—the 'operating system' of the intelligent age. Just as Microsoft defined the personal computer era and Apple defined the mobile internet era, OpenAI is attempting to seize the commanding heights of the next era.

04. Conclusion: Everything Hinges on Whether AI Can Truly Boost Productivity

In summary, OpenAI's trillion-dollar gamble is essentially a life-and-death struggle over whether AI can truly drive a revolution in productivity. The core of this gamble lies in its innovative 'revolving financing' and 'equity-for-procurement' models, which, although resolving funding issues in the short term by binding suppliers (such as NVIDIA and AMD), depend entirely on exponential growth in future AI demand for their long-term sustainability.

OpenAI anticipates that future revenue will cover its enormous costs, but this requires a prerequisite: AI must deeply penetrate various industries and bring about tangible and significant improvements in productivity. Only when corporate clients and ordinary users significantly enhance their work efficiency and reduce operational costs through AI applications, and are thus willing to continue paying, can a sufficiently large market scale be formed to consume the computing power supply constructed at exorbitant costs by OpenAI, thereby supporting its ever-expanding revenue expectations.

Conversely, if the productivity improvements brought about by AI technology fall short of expectations and market willingness to pay and user growth slow down, the current capital cycle chain supported by high expectations will break. At that point, OpenAI will not only fail to fulfill its trillion-dollar order commitments but may also trigger a collapse of the capital bubble across the entire AI industry. Therefore, whether AI can achieve a true productivity revolution directly determines whether this gamble will lay the cornerstone for the next industrial revolution or be a bubble that pre-empts the future.

- End -