Meta and Scale AI fall out! Executives bought for $14.3 billion run away, and business cooperation also fails to continue

More than two months ago, Meta spent $14.3 billion to acquire a 49% stake in Scale AI.

How long has it been since the news of discord between the two companies surfaced?

According to TechCrunch, the two parties are currently entangled in a series of issues regarding team integration and business cooperation:

• An important executive who followed Alexandr Wang (former CEO of Scale AI) to Meta has quickly left;

• There are constant frictions between Meta's old employees and those who came from Scale AI;

• Meta's internal researchers complain that the data quality from Scale AI is too low and have adopted data from Scale AI's competitors;

• ……

To be honest, while the emergence of these frictions is not surprising, the contrast to the previous "glory" is still quite noticeable.

Before the collaboration, Scale AI had become one of the highest-grossing AI startups, and its founder was even invited to serve as CEO of OpenAI during the internal power struggles at OpenAI.

On Meta's side, despite the turmoil caused by the performance of Llama 4, Mark Zuckerberg certainly has no shortage of cash. Thus, Meta made a stunning acquisition of a stake in Scale AI at a price that shocked Silicon Valley, and both parties rose to the top of trending discussions, enjoying a moment of glory.

However, it seems that both sides are not having a good time now.

After losing its CEO and some employees, Scale AI has undergone large-scale layoffs and has also lost major clients like OpenAI and Google.

As for Meta, it has undergone four reorganizations of its AI department within six months, and the high-priced hires from earlier have already started to leave, causing even Meta's old employees to feel uneasy……

As a result, when news of discord between the two parties broke, netizens began to speculate:

Mark Zuckerberg has paid for a lesson.

As for what exactly happened, here are the details—

Frictions in Personnel and Business

Since the weekend, news of discord between the two parties has been continuously fermenting, becoming a hot topic of discussion.

In summary, the discord mainly manifests in personnel and business aspects.

The first issue is that the executive who originally joined Meta alongside Alexandr Wang—Ruben Mayer—has now left

He was originally the Senior Vice President of GenAI Product Operations at Scale AI, brought in by Alexandr Wang to help operate the Superintelligent Lab.

However, later reports indicated that during his tenure at Meta, he was mainly responsible for the data operations team, rather than the core department of the Superintelligent Lab, the TBD Lab.

This means he may have been excluded from Meta's core departments.

Moreover, it wasn't just Ruben Mayer; reports also vaguely mentioned that several other executives who moved from Scale AI to Meta were unable to enter the TBD Lab.

As a result, claims began to emerge that there might be a rift in team integration between the original Scale AI team and Meta.

Even though it was reported that Ruben Mayer recently refuted these claims—according to him, he has been a part of the TBD Lab since day one and left only for personal reasons—people still have not changed this perception.

In addition, a larger conflict lies at the business level.

According to five insiders who spoke to TechCrunch, the TBD Lab is collaborating with third-party data labeling suppliers outside of Scale AI to train its upcoming AI models, including Scale AI's two biggest competitors, Mercor and Surge.

Although this is not uncommon (it is said that Meta had partnerships with two other companies before establishing the TBD Lab), this situation is somewhat different—

After all, Meta later spent heavily on Scale AI, considering them part of the family, yet the "fertile water actually flowed to outsiders."

Moreover, Meta has reportedly been quite disrespectful, with internal researchers complaining about the poor data quality from Scale AI.

At the time, the outside world generally interpreted Meta's move to "package" Alexandr Wang and several executives as more of a "talent acquisition" rather than purely focusing on Scale AI's data business.

But now it seems they haven't gained anything—neither did they successfully create a strong alliance to build the Superintelligent Lab as envisioned, nor did they leverage the data value of Scale AI itself.

In just over two months, Meta has become a complete mess.

First, Meta's Mark Zuckerberg has been continuously high-profile in poaching talent, and after bringing them in, there have been multiple reorganizations of the AI department causing internal turmoil, followed by a series of reports about employees leaving.

Not only have they poached from OpenAI (even the new Chief Scientist Zhao Shengjia has been exposed as having attempted to return to OpenAI), but even Meta's originally core employees have been fleeing As for Scale AI, although it recently secured a $99 million government contract, there are still many issues.

Shortly after its CEO and some employees joined Meta, the company underwent a major layoff (cutting 200 employees, about 14% of its total workforce) and severed ties with 500 global contractors.

Later, due to its competitive relationship with Meta, former major clients like OpenAI and Google also announced they would stop cooperating with Scale AI.

From the current perspective, the subsequent developments for both parties in this cooperation can be described as unsatisfactory.

One More Thing

By the way, it has been reported that Meta plans to launch a new model before the end of the year.

Considering the negative impact brought by Llama 4, it is said that Meta internally discussed using models from Google or OpenAI to support its own social media applications.

In other words, Meta is willing to leverage the power of other competitors to restore its reputation.

Once the news broke, netizens criticized: So the expensive hires are just calling APIs?

In fact, using external models is not an isolated incident for Meta; some netizens pointed out its previous collaboration with Midjourney.

About two weeks ago, Alexandr Wang announced that he had established a partnership with Midjourney to apply the company's technology in Meta's future models and technologies.

As the current Chief AI Officer of Meta, he provided the following reason:

Everything is to ensure that Meta can provide the best products for people.

Moreover, not only Midjourney, but even earlier, a Meta engineer revealed that the internal development at Meta has already replaced its own Llama with Claude Sonnet for coding.

Fine, netizens' spicy comments may be late but are here:

Author of this article: Yi Shui, Source: Quantum Bits, Original Title: "Meta and Scale AI Fall Out! Executives Bought for 14.3 Billion Leave, Business Cooperation Also Fails"

Risk Warning and Disclaimer

The market has risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial conditions, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investing based on this is at your own risk