What divine company prompted OpenAI to invest 3.6 billion yuan in its acquisition?

Author: Xi Wen from Pencil Dao

OpenAI is on a buying spree again.

Recently, it spent 3.6 billion yuan to acquire a startup company: Rockset. This amount broke OpenAI's acquisition record.

Founded in 2016, Rockset achieved an exit in just 8 years, truly a fortune beyond imagination.

Rockset has a team size of approximately 51-200 people, with around 86 users associated with it according to LinkedIn data.

If we calculate based on 80 people, the acquisition price of 3.6 billion yuan means that each person is valued at 45 million yuan.

Rockset was born in a small office in Silicon Valley.

In 2016, Venkat left Facebook, resigning from his eight-year position as Engineering Director of the Infrastructure Team. There, he managed online data for 1.5 billion users.

According to Pencil Dao DATA, Rockset has raised a total of 3 rounds of funding in the past 8 years, with a total amount exceeding 100 million US dollars, including investors such as Sequoia Capital.

Its core product is a "real-time analytics database," which incorporates a core search technology.

For example, when a user wants to view real-time summaries and athlete information from NBA games, Rockset can quickly provide the latest data.

So, the question arises: Why did OpenAI spend 3.6 billion yuan (500 million US dollars) to acquire such a startup company?

This can be seen from its acquisition history.

In the past year, OpenAI has not made many acquisitions: there have been approximately 3 public ones.

One occurred on August 16, 2023: Global Illumination was acquired; the next two both occurred in June 2024 - one was OpenAI's acquisition of Multi, and the other was Rockset mentioned at the beginning of this article.

In the first case, Global Illumination is an AI application company whose core product is "creative tools": using AI to help users with creative design.

After the acquisition, the entire team of Global Illumination joined OpenAI and shifted to developing the latter's core products, such as ChatGPT.

On another level, Global Illumination's products also began to integrate OpenAI's genes.

For example, Global Illumination has a game product called Biomes, where NPC conversations utilize ChatGPT.

In the second case, Multi is a remote collaboration company that provides "remote collaboration tools." After the acquisition, most members of Multi joined OpenAI.

In addition, Multi's collaboration technology was also integrated to enhance the PC version experience of ChatGPT - making team collaboration and communication possible.

Most surprisingly, after the acquisition, Multi discontinued its own products: the CEO announced that after July 24th of this year, the product will be shut down, new user registrations will not be allowed, and all user data will be deleted.

From the above two cases, we can see that OpenAI's acquisitions usually serve two main purposes.

First is talent acquisition. There is a consensus in the startup circle that people come before things. When solving any problem, people come before the task.

In the current AI industry, there is a huge talent gap. Taking China as an example, according to data from Fudan University and the Greater Bay Area AI Digital Industry Alliance, the AI talent gap is between 5 million and 10 million, or even more.

The United States also faces this issue.

Take Rockset as an example, after the acquisition, the team will also become a part of OpenAI. The team's genes mainly come from companies such as Facebook, Yahoo, Google, etc.

The second is technology integration. After this acquisition of Rockset, its technology will be integrated into OpenAI's existing products to enhance data retrieval and analysis capabilities.

Of course, from a long-term perspective, it is not excluded that OpenAI has the potential to make a major push into applications. In the AI industry, apart from hardware facilities, large models are the biggest prize. Combined with its business model characteristics, there is a possibility that one company could dominate the industry.

However, different from the application segment, its decentralized characteristics are more pronounced. While OpenAI cannot deploy in all areas, it can consider it as a fertile soil to cultivate seeds (application companies) and fertilize them (with OpenAI resources) to help them grow.

This is a meaningful long-term strategic plan that can not only enhance OpenAI's ecological competitive barriers but also lay a solid foundation for its subsequent second curve, expanding its business empire.