Mark Zuckerberg offered him $1.4 billion to work elsewhere, and Tim Cook didn't even attempt to persuade him to stay.

Friends, it's not that Apple's AI has no hope, nor is it that Cook doesn't care about talents...

What more could you do to show more sincerity than offering a compensation package worth 1.4 billion RMB?

Yes, Pang Ruoming, the head of Apple's foundation model team and an alumnus of Shanghai Jiao Tong University, gave up leading a team of over 100 people at Apple and joined Meta's super lab with a team of over 10 people. Part of the reason has been exposed.

It is said that the compensation package Zuckerberg offered - including salary, stock options - totaled up to $200 million - which is equivalent to about 1.4 billion RMB.

What does 1.4 billion RMB mean?

Within Apple, only Cook can match this level of compensation. If you add up the salaries and stock options of other executives and average them out over the years, it's really not that high. Looking at domestic listed companies, there are almost no executives or even CEOs and chairmen who can reach this level. To achieve this as an employee, you first have to thank your personal talent, and then you have to sigh about how crazy the AI era is and how crazy Zuckerberg is.

It is said that Apple didn't even attempt to match the external offer...

However, it should be noted that this $200 million compensation package is not an annual salary or a basic salary. It is an overall compensation package that includes the length of service, with a basic salary, signing bonus, and stock options that are unlocked over the years.

Before Pang Ruoming, well - known AI talents like Yu Jiahui who were poached from OpenAI are said to have received a $300 million package over 4 years, with $100 million unlocked in the first year.

Moreover, Meta and Zuckerberg don't give "thinking time" when poaching top - tier AI talents. It is said that they just name the figure and then give a "time limit for consideration", making it seem like a timed auction. They aim for a "passionate signing".

Pang Ruoming with a 1.4 - billion - RMB Compensation

Last time we mentioned that it was a bit of a surprise that Pang Ruoming, the head of Apple's foundation model team and an alumnus of Shanghai Jiao Tong University, was poached by Zuckerberg.

After all, Apple is not OpenAI. Pang Ruoming was a young executive leading a team of over 100 people at Apple, a well - respected company. He had a bright future and good prospects for money. How could he suddenly be snatched away by Meta?

Is it because there are too many old friends with the same aspirations in the super - intelligence experiment, or is he really hopeless about Apple under Cook's leadership?

Bloomberg's latest explanation is that Meta offered too much.

According to the latest revelation, Pang Ruoming's compensation package reached an almost sky - high $200 million. Even within Apple, except for CEO Cook, no other executive can come close.

Of course, this compensation is not outrageous in Meta's newly established "Super - Intelligence Lab". Because this lab, personally assembled by Zuckerberg, was previously exposed to offer $300 million over 4 years. Some outstanding talents like Yu Jiahui, an alumnus of the Juvenile Class of the University of Science and Technology of China and from OpenAI, were even offered $100 million a year.

Moreover, for this lab, Zuckerberg even acquired a 49% stake in Scale AI, a data - annotation company founded by 95 - post Chinese Alexandr Wang, for $14.3 billion, giving Alexandr Wang an income of billions of dollars apart from his salary.

How many years did Pang Ruoming sign for? We don't know.

But most contracts in this super - intelligence lab are for 4 years. So this "$200 million compensation" may be linked to the length of service, and we also have to see how Meta's stock price and market value will develop in the future.

It is said that achieving the KPIs is also an important part of this compensation structure, and he needs to help Zuckerberg achieve new growth for the company.

At least he has to pull Llama out of its current slump, right?

And to be honest, with this blatant "buying bones with a thousand pieces of gold" behavior, Zuckerberg and Meta's reputation for gambling big on AI has spread. These researchers will probably feel on tenterhooks after joining.

The "hundred - million - dollar gentlemen" have to do something extraordinary, right?

After all, the million - yuan - annual - salary genius - youth program of a domestic company back then also made many senior employees in the company who were underpaid compared to the new hires resentful. The internal letter of OpenAI's CEO later also expressed that if Meta keeps doing this, its internal culture will be in chaos and it will surely fail without a fight.

So human nature is similar, and there is not much fundamental difference across the Pacific.

After all, the salary level in Meta's super - intelligence lab is really far higher than the normal levels of other engineers at Meta and other big tech companies in Silicon Valley.

Salary Levels of Meta and Google Engineers?

The latest exposed federal documents of Meta have revealed the annual salary levels of "basic" positions such as AI research scientists, software engineers, and product managers - not including stock options, signing fees, and other benefits.

The conclusion is that among various positions at Meta, even for software engineers, the position with the highest salary ceiling, the annual basic salary is about $480,000 (equivalent to about 3.44 million RMB) at most, and at least $120,000 (equivalent to about 860,000 RMB).

Of course, the machine - learning engineer position is also not bad, with an annual salary of up to $440,000 (equivalent to about 3.15 million RMB).

For other positions, such as AI research scientists: $170,000 - $230,000, AI product marketing managers: $220,000, ML research scientists: $230,000...

In addition, data - analysis positions, which are as popular as AI, have a salary ceiling of about over $200,000. Among them, data scientists and data engineers have an annual salary of about $270,000, and ordinary data analysts have an annual salary of about $168,000 - $204,000.

Overall, before the establishment of the super - intelligence lab, Meta's compensation was in line with the high - salary levels of big tech companies in Silicon Valley.

Meanwhile, Google's salary situation has also been exposed through federal documents.

The data shows that the annual salary of thousands of its software engineers is about $340,000.

For positions such as designers, product managers, and business - direction analysts, the annual salary is about $200,000.

Research scientists have a higher salary. They have the highest income among employees, and the average upper limit of other positions is basically their lower limit.

So, to sum up, the basic annual salary of big tech companies in Silicon Valley, excluding performance - based pay, bonuses, and stock options, is basically around $300,000 (equivalent to about 2.1 million RMB).

Compared with other places, it is really high and generous.

Compared with the "annual income of $100 million" in Meta's super - intelligence lab, the difference is quite huge.

Although the "hundred - million - dollar gentlemen" that Zuckerberg has poached are truly outstanding individuals. For example, Pang Ruoming is really extraordinary.

Pang Ruoming: Leading Apple's 100 - Person Foundation Model Team

Pang Ruoming graduated from Shanghai Jiao Tong University with a bachelor's degree in 1998 and then pursued his master's and doctoral degrees at the University of Southern California and Princeton University respectively.

Right after graduation, he joined Google. After working there for 15 years, he was poached by Apple with the title of Chief Software Engineer.

During his time at Google, he was involved in and responsible for almost all important work, from early large - scale data systems, to core permission - control platforms, to deep - learning frameworks and voice models.

In the early stage, he was the co - founder and co - technical leader of the Bigtable Index and ZipIt projects.

The core of these two projects was to build a powerful structured search ability for Google. It is said that over 1000 projects within Google are using the relevant systems.

After that, he co - founded Google's Global Unified Authorization System Zanzibar with Abhishek Parmar and Zhifeng Chen (an alumnus of Fudan University).

Moreover, since 2014, he has been the sole technical leader of this project and has made the system reliability of Zanzibar reach the high - availability standard of "five nines" (99.999%), providing key support for Google's large - scale services.

The relevant paper was also included in the 2019 USENIX Annual Technical Conference.

Since 2017, he has been one of the main leaders of Google Brain's voice - recognition research.

He, Wu Yonghui, and Zhifeng Chen jointly led the development of the Babelfish/Lingvo framework, which is currently the deep - learning framework with the highest Google TPU usage rate.

Judging from the evaluation of him by Liangliang Cao (former chief scientist at Apple), the head of Google Gemini, Pang Ruoming is not only proficient in machine learning but also understands infrastructure.

Perhaps this is also the reason why Apple recruited Pang Ruoming.

In 2021, Apple poached Pang Ruoming from Google, and he has been leading Apple's foundation model team in the AI/ML department. So far, the team has about 100 people.

This team is mainly responsible for developing the core foundation models that support "Apple Intelligence" and other AI functions on Apple devices.

For example, the main dialogue function of Apple Intelligence that was unveiled at the WWDC conference is supported by the models developed by this team.

It can be said that as a Chinese engineer, the team he leads almost supports Apple's core AI strategy.

Google Scholar shows that the total citation count of his papers is over 40,000. The most - cited paper is the new - generation lightweight neural - network model MobileNetV3, which was proposed during his time at Apple.

MobileNetV3 is mainly used for image recognition and is optimized for mobile devices such as mobile phones. It can balance running speed and recognition accuracy. Since its release in 2019, it has had a profound impact in both academia and industry.

So, looking at Pang Ruoming's resume, isn't the $200 million compensation package reasonable?

Moreover, people like Pang Ruoming and Yu Jiahui are relatively simple - minded...

Otherwise, instead of signing the employment contract, a group of three or five people could form a startup company, create a few demos, and then sell them to Silicon Valley giants like Zuckerberg and Huang Renxun who have plenty of money and stocks. The income would be much higher than the current fixed salary, and they wouldn't have to feel on tenterhooks within the company after joining.

One more thing

Maybe some people may ask, why hasn't any big shot from Google DeepMind joined Meta even though Zuckerberg is poaching so aggressively?

Is it that DeepMind has such high aspirations that it looks down on Meta, or is Google's compensation also in the hundreds of millions of dollars?

Maybe neither is the most direct reason.

Because the latest statement is that, different from those Silicon Valley companies, DeepMind, which is also based in the UK, has strict non - competition and non - solicitation measures.

If you leave and still want to work in AI, be prepared to pay a sky - high liquidated damages.

Reference links:

[1]https://www.bloomberg.com/news/articles/2025 - 07 - 09/meta - poached - apple - s - pang - with - pay - package - over - 200 - million?srnd=phx - technology

[2]https://scholar.google.com/citations?user=1fsmwB8AAAAJ&hl=en

[3]https://www.businessinsider.com/meta - salaries - what - it - pays - ai - engineers - researchers - compensation - 2025 - 7

This article is from the WeChat official account "Quantum Bit" (ID: QbitAI). Author: Focusing on cutting - edge technology. Republished by 36Kr with authorization