Summary of Key Points
OpenAI’s recent revenue growth has exceeded expectations (annualized revenue in July surpassed that of the second quarter), mainly driven by the GPT-5.6 model, ChatGPT Work for businesses, and the Codex coding tool. However, it faces three major challenges:
1. It needs to justify its valuation of $85.2 billion (in preparation for an IPO).
2. It competes with Anthropic (whose valuation once exceeded OpenAI’s and whose annualized revenue reached $47 billion) and Chinese low-cost open-source models (such as Kimi K3 from Yuezhiànmiàn, which performs similarly).
3. In the future, it will require billions of dollars to build data centers to support its growing computational needs.
Detailed Analysis
1. OpenAI’s Revenue Growth: Three New Products Drive Success
OpenAI’s CFO stated that the annualized recurring revenue in July (simply put: multiplying July’s revenue by 12 to get the annual amount) was higher than that of the second quarter (April–June), and the second quarter’s performance was also strong. Behind this growth are three key products:
- GPT-5.6 series models: More powerful than previous versions, they have attracted more businesses and users willing to pay for their services.
- ChatGPT Work: An AI tool designed for businesses, helping with tasks like processing internal documents and automatically responding to customers, which businesses are willing to purchase.
- Codex coding tool: An AI that assists programmers in writing code, and its usage is increasing steadily.
These products have contributed to OpenAI’s revenue growth, supporting its high valuation.
2. The $85.2 Billion Valuation: An IPO Is Crucial for Validation
OpenAI is currently valued at $85.2 billion, which is roughly equivalent to the combined market values of two Tesla companies (currently around $400 billion each) or three Meta companies (around $280 billion each). The market will inevitably question: “Why do you deserve such a high valuation?”
Therefore, OpenAI is eager to prepare for an IPO (it has already submitted its documents, but the timing of the listing is not yet known). An IPO will not only help raise funds but also allow investors to see its actual profitability and prove that its valuation is justified. If revenue continues to grow, the valuation may be more stable during the IPO; otherwise, it could face skepticism from the market.
3. Competitors: Anthropic Challenges OpenAI, and Chinese Models Compete on Price
OpenAI’s competitors fall into two categories:
- Anthropic: This company has made a strong start this year, with a valuation that once exceeded OpenAI’s. In May, it reported annualized revenue of $47 billion (originally expected to be only $10 billion by 2025), mainly due to its popular Claude Code coding tool. OpenAI’s board chairman acknowledges the need to catch up in the coding market but also notes that Codex is gaining traction, and some users of Claude Code have switched to Codex due to higher costs.
- Chinese open-source models: Models like Kimi K3 from Yuezhiànmiàn are closing the performance gap with OpenAI’s and Anthropic’s top models and, in some tests, even surpassing them. The key advantage is their lower cost—open-source models can be modified and deployed without paying high subscription fees. This poses a threat to OpenAI, as users may prefer cheaper alternatives with similar performance.
4. The Billion-Dollar Capital Gap: Building Data Centers Is Urgent
Training AI models requires numerous servers and computational power. OpenAI plans to invest $600 billion by 2030 in computing resources (purchasing chips and building data centers). Recently, it discussed a $25 billion partnership with Nvidia to lease a large AI data center in Ohio.
Why such a significant investment? As AI models become more complex, the demand for computational power grows exponentially. For example, training GPT-4 cost hundreds of millions of dollars; future versions like GPT-5 and GPT-6 will require even more resources. The funds will come from revenue and potentially through IPO financing or partnerships with companies like Nvidia.
5. The Coding Tool Market: Developers Are a Crucial Asset
Coding tools are a competitive area for AI companies, as developers can improve efficiency using AI to write code and are willing to pay for these tools. OpenAI’s Codex competes directly with Anthropic’s Claude Code:
- Anthropic is generating revenue quickly with Claude Code, but users have complained about high costs.
- OpenAI’s Codex has seen steady growth and, due to its more affordable price (or better cost-effectiveness), has attracted many users who previously used Claude Code.
The company that can retain the most developers will gain a strong position in this market and may even drive sales of other products. For example, if developers use Codex, they might also purchase OpenAI’s other models.
Conclusion
OpenAI’s situation is mixed: revenue growth provides confidence, but it must address challenges such as its high valuation, competitive pressures, and capital needs. Whether it can prove its value before the IPO and maintain a competitive edge against Anthropic and Chinese models will depend on future product developments and cost control. For consumers, these competitive efforts could result in cheaper and more user-friendly AI tools, as businesses competing on price benefit consumers.