Summary of Key Points
In August 2026, the two giants in the AI large-model space, OpenAI and Anthropic, both experienced significant personnel changes on the eve of their IPOs. However, these changes were not mere workplace disputes; they reflected the painful transition of the AI industry from a “laboratory ideal” to a “commercial reality.” OpenAI restructured its team and cut off non-revenue-generating businesses in order to prepare for the IPO, shifting its focus from consumer (C) to business (B) customers. Anthropic, on the other hand, achieved explosive growth by recruiting top talent and targeting the corporate market. Although the two companies took different commercialization paths and faced different challenges in their IPOs, they both realized the same ultimate goal: large models must rely on their ability to generate revenue, cost control, and meet the actual needs of businesses to survive.
I. Two Different Commercialization Paths: OpenAI’s Struggles vs. Anthropic’s Success
In simple terms, the two companies have completely different approaches to generating revenue:
- OpenAI: Shifting from Consumer to Business Customers
OpenAI initially gained popularity thanks to its consumer users of ChatGPT (the free and subscription-based versions). Now, it is shifting its focus to business customers. Although consumer revenue once accounted for 60% of its total income, it has now surpassed business revenue, with over 2 million corporate customers. This transition is not easy; the sales team, which was previously geared towards consumers, needs to be retrained to understand the needs of businesses. Moreover, although OpenAI has a large number of consumer users (950 million monthly active users), the revenue generated from these users is not as high as that from business customers, who are willing to pay more for AI solutions that can solve real problems, such as code writing and data processing.
- Anthropic: Focusing on Business Customers from the Start
Anthropic was founded by someone who left OpenAI due to dissatisfaction with its commercialization strategy. The company has been targeting business customers from the beginning. Its product, Claude, can automatically modify code and identify bugs for businesses, integrating directly into their work processes. This approach has led to rapid revenue growth, with annual recurring revenue (ARR) increasing from $9 billion to $65 billion in just a few months, and the company has already turned a profit.
Key Difference: OpenAI is “forced to transform,” while Anthropic has “precisely identified the right timing” for entering the business market as the core source of revenue for large models.
II. The Essence of Personnel Changes: Creating Space for Commercialization
The personnel changes at both companies were not random but were aimed at aligning with their business strategies:
- OpenAI: A Major Clean-Up Before the IPO
Twelve executives, including the COO, CRO, and head of the security team, left within half a year. The reason is simple: non-revenue-generating businesses needed to be eliminated. For example, the security and ethics teams, although important, did not contribute to revenue and were therefore disbanded. The former CEOs and product managers, who were focused on consumer services, were no longer relevant for the company’s new focus on business customers. OpenAI’s CEO, Sam Altman, adopted a pragmatic approach, concentrating all resources on profitable projects (such as the enterprise version of ChatGPT) to present a strong financial report to investors.
- Anthropic: Targeting Top Talent to Fill Critical Gaps
Anthropic recruited top professionals who can help businesses generate revenue and reduce costs, such as:
- Clive Chan, who came from OpenAI to work on chip development, with the goal of making model inference more energy-efficient, as computing power is a major expense for large models.
- Caitlin Kalinowski, who brought expertise in hardware to expand Claude’s capabilities to include robotics and intelligent devices.
- John Jumper, a Nobel laureate, to develop AI for scientific research, strengthening the company’s technological advantages.
Summary: OpenAI is cutting unnecessary expenses, while Anthropic is strengthening its core competencies to ensure commercial success.
III. The Two IPO Challenges: Growth for Anthropic, Stability for OpenAI
Both companies are preparing for IPOs, but the capital market has different criteria for evaluating them:
- Anthropic: Selling a Story of Future Revenue
The market expects Anthropic to be valued at $2 trillion, with a price-to-sales ratio (market value ÷ annual revenue) of 30 times. Investors are interested in its potential future earnings, such as projected revenue of $200 billion in 2028. As long as the revenue growth rate remains stable, the valuation will be supported. However, there are risks: AI technology is becoming more affordable, and open-source models may compete for market share. Additionally, Anthropic must pay SpaceX $15 billion annually for computing power, which is a significant expense.
- OpenAI: Hindered by Governance Issues
OpenAI’s valuation is only $85.2 billion, with a price-to-sales ratio of 21 times, lower than Anthropic’s. Investors are concerned about the company’s management stability, given the high turnover of executives and a previous board dismissal in 2023. They are questioning whether the company can maintain its operational efficiency.
In One Sentence: Anthropic is valued for its high growth potential, while OpenAI faces challenges due to its governance issues.
IV. The Future of Large Models: The Competition Lies in Cost, Efficiency, and Real Needs
Regardless of their paths, both companies must address the same question: can large models become a profitable business? The outcomes are clear:
1. The era of “random use of AI” is over. Businesses are no longer spending money on unnecessary AI applications but are calculating the value each dollar of AI investment brings. Only those with faster, more accurate, and more cost-effective models will remain in use.
2. Companies must develop their own chips, as computing power is a major expense. OpenAI has developed its own Jalapeño chip, and Anthropic has also established a chip team. As Broadcom’s CEO said, “Top model companies will eventually produce their own chips.”
3. The divide between academia and industry in AI research is becoming more pronounced. Scientists working on AGI (general artificial intelligence) are either pursuing pure research or being replaced by those with business expertise. After going public, companies focus on generating revenue, not on exploring the future.
Conclusion: The key to success in the AI industry lies in cost efficiency, model effectiveness, and meeting the real needs of businesses. The competition between OpenAI and Anthropic reflects the transformation of the entire AI industry.
Final Thought
The story of AI large models has evolved from a “science fiction concept” to a practical business endeavor. Only companies that can generate revenue, control costs, and provide practical solutions for businesses will thrive. The competition between OpenAI and Anthropic is a microcosm of this transformation.