Summary of Key Points
OpenAI has officially submitted its S-1 (pre-IPO document) to the US Securities and Exchange Commission (SEC), but emphasizes that “whether it will ultimately go public is still undecided.” At the same time, it released a lengthy essay outlining its values, declaring the vision of providing Artificial General Intelligence (AGI) to everyone on Earth. On the other hand, its competitor Anthropic secretly submitted its S-1 a week ago. Both companies are valued at nearly one trillion dollars, and it is highly likely they will exceed this figure upon going public. OpenAI, which started as a purely non-profit organization in 2015, has gradually become commercialized due to the pressure of high costs and has undergone structural reorganizations to reach this stage. Its current annual revenue exceeds $20 billion, but it also incurs substantial losses, necessitating capital from the market to support its computing power and research and development efforts.
Breakdown and Interpretation
1. IPO as a “Pre-emptive Move” Rather Than a “Mandatory Requirement”? OpenAI’s Flexible Approach
You can consider the S-1 draft as an application for listing, which must detail the company’s business, financial data, and risks. However, OpenAI’s approach is unique:
- Proactive Disclosure: Fearing leaks of the draft, OpenAI chose to make it public, demonstrating transparency.
- Flexibility: It clearly states that the decision to go public will be made later on, unlike Anthropic, which waits for SEC review and market conditions after submitting its application.
The logic behind this is that OpenAI wants to leverage the “signal effect” of the IPO (to show the market its capabilities) without being constrained by post-listing regulations such as regular financial reporting and accountability to shareholders. Given that AGI research is a long-term and costly endeavor, maintaining flexibility is crucial.
2. Altman’s Essay on Values: AGI Should Be Accessible to Everyone
Altman’s essay, “Built to Benefit Everyone,” emphasizes that AGI should not be limited to a few; it should benefit ordinary people. He uses a relatable example:
- Before electricity was available in the 1920s, people went to bed at sunset and washed clothes by hand; with electricity, nights became longer, and household chores became easier, opening up new possibilities for society.
- AI, like electricity, is valuable not for its technology itself but for what ordinary people can do with it—understanding medical bills, learning new skills, running a small business, caring for parents, etc.
Altman also highlights two key points:
- Power Distribution: AGI should not be monopolized by a few companies or governments; power should be distributed to allow more people to participate.
- AI Should Serve People: AI should not completely automate decision-making; humans should retain control, with AI serving as a tool.
He even suggests establishing an international organization to coordinate AGI research and, if necessary, slowing down progress to ensure safety and ethical considerations keep up with technological development.
3. From Non-Profit to IPO: OpenAI’s Evolution in Finance
OpenAI’s commercialization was driven by the need to manage its high costs:
- 2015: Purely non-profit, funded by donations from Sam Altman, Elon Musk, and others, with the mission of benefiting all humanity without making a profit.
- 2019: The cost of training large models (e.g., GPT-3, which cost tens of millions of dollars) became unsustainable, leading to the creation of a “profit-capped subsidiary” that could attract investments with limited returns (e.g., initially set at 100 times the investment).
- End of 2022: ChatGPT’s success led to a surge in commercial value and massive capital inflows.
- End of 2023: A power struggle within the board resulted in Altman’s removal, but he was reinstated due to pressure from employees and investors, reflecting the conflict between the non-profit mission and commercial ambitions.
- 2025: Reorganization into a “Public Benefit Corporation (PBC)” structure with a foundation holding 26% of the shares (responsible for the mission), Microsoft holding 27% (largest shareholder), and employees/investors holding 47% (responsible for business operations). This IPO is a natural outcome of this reorganization.
4. The Competition with Anthropic: Both Aim for Trillion-Dollar Valuations
Anthropics, formerly an OpenAI employee, has also made significant moves:
- Secretly submitted its S-1 a week ago.
- As of the end of May, it was valued at $96.5 billion, surpassing OpenAI’s $85.2 billion, making it the most valuable AI startup in the world.
- Both companies are close to the trillion-dollar milestone and are likely to exceed this figure upon going public.
This indicates that the AI industry has entered a phase of competitive dominance, with capital highly optimistic about the future of AGI—whichever company develops true AGI first will control the technological narrative for decades.
5. Profitability Despite High Costs: OpenAI’s Dilemma
OpenAI’s current financial situation is one of high revenue and substantial losses:
- Annual revenue exceeds $20 billion (about $2 billion per month), mainly from ChatGPT Plus subscriptions and enterprise API services.
- Expected losses in 2026 range from $14 to $25 billion due to high costs (e.g., GPU rental fees) and ongoing R&D investments.
Therefore, the IPO is not about cashing out but about securing more capital to continue funding AGI research.
Conclusion
OpenAI’s IPO is both a necessity for commercialization and a requirement for AGI development. While it promotes the ideal of serving all with AGI, it must rely on the capital market to address its financial challenges. Whether it can balance commercial interests with its non-profit mission and avoid being controlled by capital remains to be seen. The competition in the AI industry is just beginning to heat up.